remove rebase debug

TODO:
- cleanup routes interface
- internal/external states

convert/convert.go

@@ -191,8 +191,6 @@ func ConvertModel(fsys fs.FS, f *os.File) error {
 		conv = &phi3Model{}
 	case "Qwen2ForCausalLM":
 		conv = &qwen2Model{}
-	case "Qwen2_5_VLForConditionalGeneration":
-		conv = &qwen25VLModel{}
 	case "BertModel":
 		conv = &bertModel{}
 	case "CohereForCausalLM":

convert/convert_llama.go

@@ -139,8 +139,7 @@ func (p *llamaModel) Tensors(ts []Tensor) []*ggml.Tensor {
 	}
 
 	for _, t := range ts {
-		if strings.HasSuffix(t.Name(), "attn_q.weight") || strings.HasSuffix(t.Name(), "attn_k.weight") ||
-			strings.HasSuffix(t.Name(), "attn_q_proj.weight") || strings.HasSuffix(t.Name(), "attn_k_proj.weight") {
+		if strings.HasSuffix(t.Name(), "attn_q.weight") || strings.HasSuffix(t.Name(), "attn_k.weight") {
 			if !p.skipRepack {
 				t.SetRepacker(p.repack)
 			}
@@ -182,9 +181,9 @@ func (p *llamaModel) repack(name string, data []float32, shape []uint64) ([]floa
 	}
 
 	var heads uint32
-	if strings.HasSuffix(name, "attn_q.weight") || strings.HasSuffix(name, "attn_q_proj.weight") {
+	if strings.HasSuffix(name, "attn_q.weight") {
 		heads = p.NumAttentionHeads
-	} else if strings.HasSuffix(name, "attn_k.weight") || strings.HasSuffix(name, "attn_k_proj.weight") {
+	} else if strings.HasSuffix(name, "attn_k.weight") {
 		heads = cmp.Or(p.NumKeyValueHeads, p.NumAttentionHeads)
 	} else {
 		return nil, fmt.Errorf("unknown tensor for repack: %s", name)

convert/convert_qwen2.go

@@ -15,7 +15,6 @@ type qwen2Model struct {
 		Type                          string     `json:"type"`
 		Factor                        ropeFactor `json:"factor"`
 		OriginalMaxPositionEmbeddings uint32     `json:"original_max_position_embeddings"`
-		MropeSection                  []int32    `json:"mrope_section"`
 	} `json:"rope_scaling"`
 	RMSNormEPS float32 `json:"rms_norm_eps"`
 }
@@ -40,8 +39,6 @@ func (q *qwen2Model) KV(t *Tokenizer) ggml.KV {
 	case "yarn":
 		kv["qwen2.rope.scaling.type"] = q.RopeScaling.Type
 		kv["qwen2.rope.scaling.factor"] = q.RopeScaling.Factor
-	case "mrope", "default":
-		kv["qwen2.rope.mrope_section"] = q.RopeScaling.MropeSection
 	default:
 		panic("unknown rope scaling type")
 	}

convert/convert_qwen25vl.go

@@ -1,102 +0,0 @@
-package convert
-
-import (
-	"cmp"
-	"slices"
-	"strings"
-
-	"github.com/ollama/ollama/fs/ggml"
-)
-
-type qwen25VLModel struct {
-	qwen2Model
-
-	VisionModel struct {
-		Depth               uint32  `json:"depth"`
-		HiddenSize          uint32  `json:"hidden_size"`
-		NumHeads            uint32  `json:"num_heads"`
-		InChannels          uint32  `json:"in_chans"`
-		PatchSize           uint32  `json:"patch_size"`
-		SpatialMergeSize    uint32  `json:"spatial_merge_size"`
-		SpatialPatchSize    uint32  `json:"spatial_patch_size"`
-		WindowSize          uint32  `json:"window_size"`
-		RMSNormEps          float32 `json:"layer_norm_epsilon"`
-		RopeTheta           float32 `json:"rope_theta"`
-		FullAttentionBlocks []int32 `json:"fullatt_block_indexes"`
-		TemporalPatchSize   uint32  `json:"temporal_patch_size"`
-	} `json:"vision_config"`
-}
-
-var _ ModelConverter = (*qwen25VLModel)(nil)
-
-func (q *qwen25VLModel) KV(t *Tokenizer) ggml.KV {
-	kv := q.ModelParameters.KV(t)
-	kv["general.architecture"] = "qwen25vl"
-
-	for k, v := range q.qwen2Model.KV(t) {
-		if strings.HasPrefix(k, "qwen2.") {
-			kv[strings.Replace(k, "qwen2.", "qwen25vl.", 1)] = v
-		}
-	}
-
-	if q.VisionModel.FullAttentionBlocks == nil {
-		kv["qwen25vl.vision.fullatt_block_indexes"] = []int32{7, 15, 23, 31}
-	}
-
-	kv["qwen25vl.vision.block_count"] = cmp.Or(q.VisionModel.Depth, 32)
-	kv["qwen25vl.vision.embedding_length"] = q.VisionModel.HiddenSize
-	kv["qwen25vl.vision.attention.head_count"] = cmp.Or(q.VisionModel.NumHeads, 16)
-	kv["qwen25vl.vision.num_channels"] = q.VisionModel.InChannels
-	kv["qwen25vl.vision.patch_size"] = cmp.Or(q.VisionModel.PatchSize, 14)
-	kv["qwen25vl.vision.spatial_merge_size"] = cmp.Or(q.VisionModel.SpatialMergeSize, 2)
-	kv["qwen25vl.vision.spatial_patch_size"] = q.VisionModel.SpatialPatchSize
-	kv["qwen25vl.vision.window_size"] = cmp.Or(q.VisionModel.WindowSize, 112)
-	kv["qwen25vl.vision.attention.layer_norm_epsilon"] = cmp.Or(q.VisionModel.RMSNormEps, 1e-6)
-	kv["qwen25vl.vision.rope.freq_base"] = cmp.Or(q.VisionModel.RopeTheta, 1e4)
-	kv["qwen25vl.vision.fullatt_block_indexes"] = q.VisionModel.FullAttentionBlocks
-	kv["qwen25vl.vision.temporal_patch_size"] = cmp.Or(q.VisionModel.TemporalPatchSize, 2)
-
-	return kv
-}
-
-func (q *qwen25VLModel) Tensors(ts []Tensor) []*ggml.Tensor {
-	var out []*ggml.Tensor
-
-	for _, t := range ts {
-		if strings.Contains(t.Name(), "patch_embed.proj") {
-			for t := range splitDim(t, 2,
-				strings.NewReplacer("patch_embed.proj", "patch_embd_0"),
-				strings.NewReplacer("patch_embed.proj", "patch_embd_1"),
-			) {
-				t.Shape = slices.DeleteFunc(t.Shape, func(i uint64) bool { return i == 1 })
-				out = append(out, t)
-			}
-		} else if strings.Contains(t.Name(), "attn.qkv") {
-			out = append(out, slices.Collect(splitDim(t, 0,
-				strings.NewReplacer("attn.qkv", "attn_q"),
-				strings.NewReplacer("attn.qkv", "attn_k"),
-				strings.NewReplacer("attn.qkv", "attn_v"),
-			))...)
-		} else {
-			out = append(out, &ggml.Tensor{
-				Name:     t.Name(),
-				Kind:     t.Kind(),
-				Shape:    t.Shape(),
-				WriterTo: t,
-			})
-		}
-	}
-
-	return out
-}
-
-func (p *qwen25VLModel) Replacements() []string {
-	return append(
-		p.qwen2Model.Replacements(),
-		"visual", "v",
-		"blocks", "blk",
-		"attn.proj", "attn_out",
-		"norm1", "ln1",
-		"norm2", "ln2",
-	)
-}

convert/tensor.go

@@ -1,56 +0,0 @@
-package convert
-
-import (
-	"iter"
-	"slices"
-	"strings"
-
-	"github.com/ollama/ollama/fs/ggml"
-	"github.com/pdevine/tensor"
-	"github.com/pdevine/tensor/native"
-)
-
-// splitDim splits a tensor along a specified dimension into multiple tensors. The dimension
-// is split evenly based on the number of replacers provided.
-func splitDim(t Tensor, dim int, replacers ...*strings.Replacer) iter.Seq[*ggml.Tensor] {
-	return func(yield func(*ggml.Tensor) bool) {
-		for i, replacer := range replacers {
-			shape := slices.Clone(t.Shape())
-			shape[dim] = shape[dim] / uint64(len(replacers))
-
-			slice := slices.Repeat([]tensor.Slice{nil}, len(shape))
-			slice[dim] = tensor.S(i*int(shape[dim]), (i+1)*int(shape[dim]))
-
-			tt := t.Clone()
-			tt.SetRepacker(func(_ string, data []float32, shape []uint64) ([]float32, error) {
-				dims := make([]int, len(shape))
-				for i := range shape {
-					dims[i] = int(shape[i])
-				}
-
-				var t tensor.Tensor = tensor.New(tensor.WithShape(dims...), tensor.WithBacking(data))
-				t, err := t.Slice(slice...)
-				if err != nil {
-					return nil, err
-				}
-
-				t = tensor.Materialize(t)
-				// flatten tensor so it can be written as a vector
-				if err := t.Reshape(t.Shape().TotalSize()); err != nil {
-					return nil, err
-				}
-
-				return native.VectorF32(t.(*tensor.Dense))
-			})
-
-			if !yield(&ggml.Tensor{
-				Name:     replacer.Replace(t.Name()),
-				Kind:     t.Kind(),
-				Shape:    shape,
-				WriterTo: tt,
-			}) {
-				break
-			}
-		}
-	}
-}

fs/ggml/ggml.go

@@ -126,7 +126,6 @@ func (kv KV) OllamaEngineRequired() bool {
 		"mistral3",
 		"llama4",
 		"mllama",
-		"qwen25vl",
 	}, kv.Architecture())
 }
 
@@ -650,20 +649,6 @@ func (llm GGML) VisionGraphSize() (weights, graphSize uint64) {
 		graphSize = 4 * (imageSize*imageSize*numChannels +
 			embeddingLength*patchSize +
 			numPatches*numPatches*headCount)
-	case "qwen25vl":
-		maxPixels := uint64(llm.KV().Uint("vision.max_pixels", 28*28*1280))
-
-		numPatches := maxPixels / (patchSize * patchSize)
-
-		graphSize = 4 * (maxPixels*numChannels + // Original image storage
-			// Normalized pixels
-			maxPixels*numChannels +
-			// Patches storage (numPatches * channels * patchSize^2)
-			numPatches*numChannels*patchSize*patchSize +
-			// Self-attention calculations
-			numPatches*numPatches*headCount +
-			// Additional buffer for processing
-			embeddingLength*numPatches)
 	case "llama4":
 		// vision graph is computed independently in the same schedule
 		// and is negligible compared to the worst case text graph

go.mod

@@ -19,6 +19,7 @@ require (
 	github.com/d4l3k/go-bfloat16 v0.0.0-20211005043715-690c3bdd05f1
 	github.com/dlclark/regexp2 v1.11.4
 	github.com/emirpasic/gods/v2 v2.0.0-alpha
+	github.com/go-json-experiment/json v0.0.0-20250417205406-170dfdcf87d1
 	github.com/google/go-cmp v0.6.0
 	github.com/mattn/go-runewidth v0.0.14
 	github.com/nlpodyssey/gopickle v0.3.0

go.sum

@@ -69,6 +69,8 @@ github.com/go-fonts/latin-modern v0.2.0/go.mod h1:rQVLdDMK+mK1xscDwsqM5J8U2jrRa3
 github.com/go-fonts/liberation v0.1.1/go.mod h1:K6qoJYypsmfVjWg8KOVDQhLc8UDgIK2HYqyqAO9z7GY=
 github.com/go-fonts/stix v0.1.0/go.mod h1:w/c1f0ldAUlJmLBvlbkvVXLAD+tAMqobIIQpmnUIzUY=
 github.com/go-gl/glfw v0.0.0-20190409004039-e6da0acd62b1/go.mod h1:vR7hzQXu2zJy9AVAgeJqvqgH9Q5CA+iKCZ2gyEVpxRU=
+github.com/go-json-experiment/json v0.0.0-20250417205406-170dfdcf87d1 h1:+VexzzkMLb1tnvpuQdGT/DicIRW7MN8ozsXqBMgp0Hk=
+github.com/go-json-experiment/json v0.0.0-20250417205406-170dfdcf87d1/go.mod h1:TiCD2a1pcmjd7YnhGH0f/zKNcCD06B029pHhzV23c2M=
 github.com/go-latex/latex v0.0.0-20210118124228-b3d85cf34e07/go.mod h1:CO1AlKB2CSIqUrmQPqA0gdRIlnLEY0gK5JGjh37zN5U=
 github.com/go-playground/assert/v2 v2.2.0 h1:JvknZsQTYeFEAhQwI4qEt9cyV5ONwRHC+lYKSsYSR8s=
 github.com/go-playground/assert/v2 v2.2.0/go.mod h1:VDjEfimB/XKnb+ZQfWdccd7VUvScMdVu0Titje2rxJ4=

llama/patches/0015-add-argsort-and-cuda-copy-for-i32.patch

@@ -1,277 +0,0 @@
-From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
-From: Michael Yang <git@mxy.ng>
-Date: Thu, 1 May 2025 13:45:12 -0700
-Subject: [PATCH] add argsort and cuda copy for i32
-
----
- ggml/src/ggml-cpu/ops.cpp     |  43 ++++++++++++++
- ggml/src/ggml-cuda/argsort.cu | 102 +++++++++++++++++++++++++++++++++-
- ggml/src/ggml-cuda/cpy.cu     |  49 ++++++++++++++++
- 3 files changed, 192 insertions(+), 2 deletions(-)
-
-diff --git a/ggml/src/ggml-cpu/ops.cpp b/ggml/src/ggml-cpu/ops.cpp
-index becdae07..7a44b6cf 100644
---- a/ggml/src/ggml-cpu/ops.cpp
-+++ b/ggml/src/ggml-cpu/ops.cpp
-@@ -6890,6 +6890,45 @@ static void ggml_compute_forward_argsort_f32(
-     }
- }
- 
-+static void ggml_compute_forward_argsort_i32(
-+    const ggml_compute_params * params,
-+    ggml_tensor * dst) {
-+
-+    const ggml_tensor * src0 = dst->src[0];
-+
-+    GGML_TENSOR_UNARY_OP_LOCALS
-+
-+    GGML_ASSERT(nb0 == sizeof(int32_t));
-+
-+    const int ith = params->ith;
-+    const int nth = params->nth;
-+
-+    const int64_t nr = ggml_nrows(src0);
-+
-+    ggml_sort_order order = (ggml_sort_order) ggml_get_op_params_i32(dst, 0);
-+
-+    for (int64_t i = ith; i < nr; i += nth) {
-+        int32_t * dst_data = (int32_t *)((char *) dst->data + i*nb1);
-+        const int32_t * src_data = (int32_t *)((char *) src0->data + i*nb01);
-+
-+        for (int64_t j = 0; j < ne0; j++) {
-+            dst_data[j] = j;
-+        }
-+
-+        // C doesn't have a functional sort, so we do a bubble sort instead
-+        for (int64_t j = 0; j < ne0; j++) {
-+            for (int64_t k = j + 1; k < ne0; k++) {
-+                if ((order == GGML_SORT_ORDER_ASC  && src_data[dst_data[j]] > src_data[dst_data[k]]) ||
-+                    (order == GGML_SORT_ORDER_DESC && src_data[dst_data[j]] < src_data[dst_data[k]])) {
-+                    int32_t tmp = dst_data[j];
-+                    dst_data[j] = dst_data[k];
-+                    dst_data[k] = tmp;
-+                }
-+            }
-+        }
-+    }
-+}
-+
- void ggml_compute_forward_argsort(
-     const ggml_compute_params * params,
-     ggml_tensor * dst) {
-@@ -6901,6 +6940,10 @@ void ggml_compute_forward_argsort(
-             {
-                 ggml_compute_forward_argsort_f32(params, dst);
-             } break;
-+        case GGML_TYPE_I32:
-+            {
-+                ggml_compute_forward_argsort_i32(params, dst);
-+            } break;
-         default:
-             {
-                 GGML_ABORT("fatal error");
-diff --git a/ggml/src/ggml-cuda/argsort.cu b/ggml/src/ggml-cuda/argsort.cu
-index 607ded85..53b02634 100644
---- a/ggml/src/ggml-cuda/argsort.cu
-+++ b/ggml/src/ggml-cuda/argsort.cu
-@@ -85,13 +85,107 @@ static void argsort_f32_i32_cuda(const float * x, int * dst, const int ncols, co
-     }
- }
- 
-+
-+template<ggml_sort_order order>
-+static __global__ void k_argsort_i32_i32(const int32_t * x, int * dst, const int ncols, const int ncols_pad) {
-+    extern __shared__ int shared_mem[];
-+    int * indices = shared_mem;
-+
-+    const int tid = threadIdx.x;
-+    const int row = blockIdx.y;
-+
-+    // Initialize all indices, handling the case where threads < ncols_pad
-+    for (int i = tid; i < ncols_pad; i += blockDim.x) {
-+        indices[i] = i < ncols ? i : 0; // Use 0 for padding indices
-+    }
-+    __syncthreads();
-+
-+    // Bitonic sort
-+    for (int k = 2; k <= ncols_pad; k *= 2) {
-+        for (int j = k/2; j > 0; j /= 2) {
-+            for (int i = tid; i < ncols_pad; i += blockDim.x) {
-+                const int ij = i ^ j;
-+                if (ij > i) {
-+                    // Only compare values within the actual data range
-+                    if (i < ncols && ij < ncols) {
-+                        if ((i & k) == 0) {
-+                            if (order == GGML_SORT_ORDER_ASC) {
-+                                if (x[row * ncols + indices[i]] > x[row * ncols + indices[ij]]) {
-+                                    int tmp = indices[i];
-+                                    indices[i] = indices[ij];
-+                                    indices[ij] = tmp;
-+                                }
-+                            } else {
-+                                if (x[row * ncols + indices[i]] < x[row * ncols + indices[ij]]) {
-+                                    int tmp = indices[i];
-+                                    indices[i] = indices[ij];
-+                                    indices[ij] = tmp;
-+                                }
-+                            }
-+                        } else {
-+                            if (order == GGML_SORT_ORDER_ASC) {
-+                                if (x[row * ncols + indices[i]] < x[row * ncols + indices[ij]]) {
-+                                    int tmp = indices[i];
-+                                    indices[i] = indices[ij];
-+                                    indices[ij] = tmp;
-+                                }
-+                            } else {
-+                                if (x[row * ncols + indices[i]] > x[row * ncols + indices[ij]]) {
-+                                    int tmp = indices[i];
-+                                    indices[i] = indices[ij];
-+                                    indices[ij] = tmp;
-+                                }
-+                            }
-+                        }
-+                    }
-+                }
-+            }
-+            __syncthreads();
-+        }
-+    }
-+
-+    // Write sorted indices to output, only threads handling valid data
-+    for (int i = tid; i < ncols; i += blockDim.x) {
-+        dst[row * ncols + i] = indices[i];
-+    }
-+}
-+
-+static void argsort_i32_i32_cuda(const int32_t * x, int * dst, const int ncols, const int nrows, ggml_sort_order order, cudaStream_t stream) {
-+    // Bitonic sort requires ncols to be power of 2
-+    const int ncols_pad = next_power_of_2(ncols);
-+
-+    // Ensure thread count doesn't exceed maximum (typically 1024)
-+    const int max_threads = 1024;  // This is the typical max for most GPUs
-+    const int threads_per_block = ncols_pad > max_threads ? max_threads : ncols_pad;
-+
-+    const dim3 block_dims(threads_per_block, 1, 1);
-+    const dim3 block_nums(1, nrows, 1);
-+    const size_t shared_mem = ncols_pad * sizeof(int);
-+
-+    // Check if shared memory size is within limits
-+    const size_t max_shared_mem = ggml_cuda_info().devices[ggml_cuda_get_device()].smpb;
-+
-+    // Instead of logging an error, use GGML_ASSERT with a descriptive message
-+    GGML_ASSERT(shared_mem <= max_shared_mem && "argsort: required shared memory exceeds device limit");
-+
-+    // Launch kernels with the updated thread configuration
-+    if (order == GGML_SORT_ORDER_ASC) {
-+        k_argsort_i32_i32<GGML_SORT_ORDER_ASC><<<block_nums, block_dims, shared_mem, stream>>>(x, dst, ncols, ncols_pad);
-+    } else if (order == GGML_SORT_ORDER_DESC) {
-+        k_argsort_i32_i32<GGML_SORT_ORDER_DESC><<<block_nums, block_dims, shared_mem, stream>>>(x, dst, ncols, ncols_pad);
-+    } else {
-+        GGML_ABORT("fatal error");
-+    }
-+}
-+
-+
- void ggml_cuda_op_argsort(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
-     const ggml_tensor * src0 = dst->src[0];
-     const float * src0_d = (const float *)src0->data;
-     float * dst_d = (float *)dst->data;
-     cudaStream_t stream = ctx.stream();
- 
--    GGML_ASSERT(src0->type == GGML_TYPE_F32);
-+    GGML_ASSERT(src0->type == GGML_TYPE_F32 || src0->type == GGML_TYPE_I32);
-     GGML_ASSERT( dst->type == GGML_TYPE_I32);
-     GGML_ASSERT(ggml_is_contiguous(src0));
- 
-@@ -100,5 +194,9 @@ void ggml_cuda_op_argsort(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
- 
-     enum ggml_sort_order order = (enum ggml_sort_order) dst->op_params[0];
- 
--    argsort_f32_i32_cuda(src0_d, (int *)dst_d, ncols, nrows, order, stream);
-+    if (src0->type == GGML_TYPE_I32) {
-+        argsort_i32_i32_cuda((const int32_t *)src0_d, (int *)dst_d, ncols, nrows, order, stream);
-+    } else {
-+        argsort_f32_i32_cuda(src0_d, (int *)dst_d, ncols, nrows, order, stream);
-+    }
- }
-diff --git a/ggml/src/ggml-cuda/cpy.cu b/ggml/src/ggml-cuda/cpy.cu
-index 2d46176e..47383486 100644
---- a/ggml/src/ggml-cuda/cpy.cu
-+++ b/ggml/src/ggml-cuda/cpy.cu
-@@ -38,6 +38,13 @@ static __device__ void cpy_1_f16_f32(const char * cxi, char * cdsti) {
-     *dsti = *xi;
- }
- 
-+static __device__ void cpy_1_i32_i32(const char * cxi, char * cdsti) {
-+    const int32_t * xi = (const int32_t *) cxi;
-+    int32_t * dsti = (int32_t *) cdsti;
-+
-+    *dsti = *xi;
-+}
-+
- template <cpy_kernel_t cpy_1>
- static __global__ void cpy_f32_f16(const char * cx, char * cdst_direct, const int ne,
-                                    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
-@@ -68,6 +75,44 @@ static __global__ void cpy_f32_f16(const char * cx, char * cdst_direct, const in
-     cpy_1(cx + x_offset, cdst + dst_offset);
- }
- 
-+// First, add this template function after the other template functions
-+template <cpy_kernel_t cpy_1>
-+static __global__ void cpy_i32_i32(const char * cx, char * cdst, const int ne,
-+                                 const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
-+                                 const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11,
-+                                 const int nb12, const int nb13) {
-+    const int64_t i = blockDim.x*blockIdx.x + threadIdx.x;
-+
-+    if (i >= ne) {
-+        return;
-+    }
-+
-+    const int64_t i03 = i/(ne00 * ne01 * ne02);
-+    const int64_t i02 = (i - i03*ne00*ne01*ne02 )/ (ne00*ne01);
-+    const int64_t i01 = (i - i03*ne00*ne01*ne02  -  i02*ne01*ne00) / ne00;
-+    const int64_t i00 = i - i03*ne00*ne01*ne02 - i02*ne01*ne00 - i01*ne00;
-+    const int64_t x_offset = i00*nb00 + i01*nb01 + i02*nb02 + i03 * nb03;
-+
-+    const int64_t i13 = i/(ne10 * ne11 * ne12);
-+    const int64_t i12 = (i - i13*ne10*ne11*ne12) / (ne10*ne11);
-+    const int64_t i11 = (i - i13*ne10*ne11*ne12 - i12*ne10*ne11) / ne10;
-+    const int64_t i10 = i - i13*ne10*ne11*ne12 - i12*ne10*ne11 - i11*ne10;
-+    const int64_t dst_offset = i10*nb10 + i11*nb11 + i12*nb12 + i13 * nb13;
-+
-+    cpy_1(cx + x_offset, cdst + dst_offset);
-+}
-+
-+// Then modify the ggml_cpy_i32_i32_cuda function to use the new template
-+static void ggml_cpy_i32_i32_cuda(
-+    const char * cx, char * cdst, const int ne,
-+    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
-+    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream, char ** cdst_indirect, int graph_cpynode_index) {
-+
-+    const int num_blocks = (ne + CUDA_CPY_BLOCK_SIZE - 1) / CUDA_CPY_BLOCK_SIZE;
-+    cpy_i32_i32<cpy_1_i32_i32><<<num_blocks, CUDA_CPY_BLOCK_SIZE, 0, stream>>>
-+        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
-+}
-+
- static __device__ void cpy_blck_f32_q8_0(const char * cxi, char * cdsti) {
-     const float * xi = (const float *) cxi;
-     block_q8_0 * dsti = (block_q8_0 *) cdsti;
-@@ -631,6 +676,8 @@ void ggml_cuda_cpy(ggml_backend_cuda_context & ctx, const ggml_tensor * src0, gg
-         ggml_cpy_f16_f16_cuda (src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream, dest_ptrs_d, graph_cpynode_index);
-     } else if (src0->type == GGML_TYPE_F16 && src1->type == GGML_TYPE_F32) {
-         ggml_cpy_f16_f32_cuda (src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream, dest_ptrs_d, graph_cpynode_index);
-+    } else if (src0->type == GGML_TYPE_I32 && src1->type == GGML_TYPE_I32) {
-+        ggml_cpy_i32_i32_cuda(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream, dest_ptrs_d, graph_cpynode_index);
-     } else {
-         GGML_ABORT("%s: unsupported type combination (%s to %s)\n", __func__,
-                 ggml_type_name(src0->type), ggml_type_name(src1->type));
-@@ -686,6 +733,8 @@ void* ggml_cuda_cpy_fn(const ggml_tensor * src0, ggml_tensor * src1) {
-         return (void*) cpy_f32_f16<cpy_1_f32_f16>;
-     } else if (src0->type == GGML_TYPE_F16 && src1->type == GGML_TYPE_F32) {
-         return (void*) cpy_f32_f16<cpy_1_f16_f32>;
-+    } else if (src0->type == GGML_TYPE_I32 && src1->type == GGML_TYPE_I32) {
-+        return (void*) cpy_i32_i32<cpy_1_i32_i32>;
-     } else {
-         GGML_ABORT("%s: unsupported type combination (%s to %s)\n", __func__,
-                 ggml_type_name(src0->type), ggml_type_name(src1->type));

llm/memory.go

@@ -415,7 +415,7 @@ func projectorMemoryRequirements(filename string) (weights uint64) {
 	}
 	defer file.Close()
 
-	ggml, _, err := ggml.Decode(file, 0)
+	ggml, _, err := ggml.Decode(file, 1024)
 	if err != nil {
 		return 0
 	}

ml/backend.go

@@ -119,21 +119,6 @@ type Context interface {
 	Layer(int) Context
 }
 
-// RopeOptions contains optional parameters for RoPE function
-type RopeOptions struct {
-	OriginalContextLen uint32
-}
-
-// RopeOption defines a function that modifies RopeOpts
-type RopeOption func(*RopeOptions)
-
-// WithContextLen sets a custom context length
-func WithContextLen(len uint32) RopeOption {
-	return func(opts *RopeOptions) {
-		opts.OriginalContextLen = len
-	}
-}
-
 type Tensor interface {
 	Dim(n int) int
 	Stride(n int) int
@@ -159,7 +144,7 @@ type Tensor interface {
 	AvgPool2D(ctx Context, k, s int, p float32) Tensor
 	Conv2D(ctx Context, weight Tensor, s0, s1, p0, p1, d0, d1 int) Tensor
 
-	RoPE(ctx Context, positionIDs, ropeFactors Tensor, dim, ropeType uint32, base, scale float32, options ...RopeOption) Tensor
+	RoPE(ctx Context, positionIDs, ropeFactors Tensor, dim, ropeType uint32, base, scale float32) Tensor
 	IM2Col(ctx Context, weight Tensor, s0, s1, p0, p1, d0, d1 int) Tensor
 
 	Sin(ctx Context) Tensor
@@ -187,7 +172,6 @@ type Tensor interface {
 	Duplicate(ctx Context) Tensor
 
 	TopK(ctx Context, k int) Tensor
-	Argsort(ctx Context) Tensor
 }
 
 // ScaledDotProductAttention implements a fused attention

ml/backend/ggml/ggml.go

@@ -915,8 +915,6 @@ func (t *Tensor) RMSNorm(ctx ml.Context, w ml.Tensor, eps float32) ml.Tensor {
 func (t *Tensor) Pad(ctx ml.Context, shape ...int) ml.Tensor {
 	if len(shape) != 4 {
 		panic("expected 4 dimensions")
-	} else if shape[3] != 0 {
-		panic("cuda does not support 4d tensors")
 	}
 
 	return &Tensor{
@@ -1062,17 +1060,7 @@ const (
 	ropeTypeVision C.int = 24
 )
 
-func (t *Tensor) RoPE(ctx ml.Context, positionIDs, ropeFactors ml.Tensor, ropeDim, ropeType uint32, ropeBase, ropeScale float32, options ...ml.RopeOption) ml.Tensor {
-	// Default options
-	opts := &ml.RopeOptions{
-		OriginalContextLen: 131072,
-	}
-
-	// Apply any provided options
-	for _, option := range options {
-		option(opts)
-	}
-
+func (t *Tensor) RoPE(ctx ml.Context, positionIDs, ropeFactors ml.Tensor, ropeDim, ropeType uint32, ropeBase, ropeScale float32) ml.Tensor {
 	if ropeFactors == nil {
 		ropeFactors = &Tensor{b: t.b}
 	}
@@ -1085,19 +1073,16 @@ func (t *Tensor) RoPE(ctx ml.Context, positionIDs, ropeFactors ml.Tensor, ropeDi
 	return &Tensor{
 		b: t.b,
 		t: C.ggml_rope_ext(
-			ctx.(*Context).ctx,
-			dequant,
-			positionIDs.(*Tensor).t,
-			ropeFactors.(*Tensor).t,
+			ctx.(*Context).ctx, dequant, positionIDs.(*Tensor).t, ropeFactors.(*Tensor).t,
 			C.int(ropeDim),
 			C.int(ropeType),
-			C.int(opts.OriginalContextLen),
+			131072, // YaRN n_ctx_train
 			C.float(ropeBase),
 			C.float(ropeScale),
-			C.float(0.0),
-			C.float(1.0),
-			C.float(32.0),
-			C.float(1.0),
+			0.,  // YaRN ext_factor
+			1.,  // YaRN attn_factor
+			32., // YaRN beta_fast
+			1.,  // YaRN beta_slow
 		),
 	}
 }
@@ -1191,10 +1176,3 @@ func (t *Tensor) TopK(ctx ml.Context, k int) ml.Tensor {
 		t: C.ggml_top_k(ctx.(*Context).ctx, t.t, C.int(k)),
 	}
 }
-
-func (t *Tensor) Argsort(ctx ml.Context) ml.Tensor {
-	return &Tensor{
-		b: t.b,
-		t: C.ggml_argsort(ctx.(*Context).ctx, t.t, C.GGML_SORT_ORDER_ASC),
-	}
-}

ml/backend/ggml/ggml/src/ggml-cpu/ops.cpp

@@ -6822,45 +6822,6 @@ static void ggml_compute_forward_argsort_f32(
     }
 }
 
-static void ggml_compute_forward_argsort_i32(
-    const ggml_compute_params * params,
-    ggml_tensor * dst) {
-
-    const ggml_tensor * src0 = dst->src[0];
-
-    GGML_TENSOR_UNARY_OP_LOCALS
-
-    GGML_ASSERT(nb0 == sizeof(int32_t));
-
-    const int ith = params->ith;
-    const int nth = params->nth;
-
-    const int64_t nr = ggml_nrows(src0);
-
-    ggml_sort_order order = (ggml_sort_order) ggml_get_op_params_i32(dst, 0);
-
-    for (int64_t i = ith; i < nr; i += nth) {
-        int32_t * dst_data = (int32_t *)((char *) dst->data + i*nb1);
-        const int32_t * src_data = (int32_t *)((char *) src0->data + i*nb01);
-
-        for (int64_t j = 0; j < ne0; j++) {
-            dst_data[j] = j;
-        }
-
-        // C doesn't have a functional sort, so we do a bubble sort instead
-        for (int64_t j = 0; j < ne0; j++) {
-            for (int64_t k = j + 1; k < ne0; k++) {
-                if ((order == GGML_SORT_ORDER_ASC  && src_data[dst_data[j]] > src_data[dst_data[k]]) ||
-                    (order == GGML_SORT_ORDER_DESC && src_data[dst_data[j]] < src_data[dst_data[k]])) {
-                    int32_t tmp = dst_data[j];
-                    dst_data[j] = dst_data[k];
-                    dst_data[k] = tmp;
-                }
-            }
-        }
-    }
-}
-
 void ggml_compute_forward_argsort(
     const ggml_compute_params * params,
     ggml_tensor * dst) {
@@ -6872,10 +6833,6 @@ void ggml_compute_forward_argsort(
             {
                 ggml_compute_forward_argsort_f32(params, dst);
             } break;
-        case GGML_TYPE_I32:
-            {
-                ggml_compute_forward_argsort_i32(params, dst);
-            } break;
         default:
             {
                 GGML_ABORT("fatal error");

ml/backend/ggml/ggml/src/ggml-cuda/argsort.cu

@@ -85,107 +85,13 @@ static void argsort_f32_i32_cuda(const float * x, int * dst, const int ncols, co
     }
 }
 
-
-template<ggml_sort_order order>
-static __global__ void k_argsort_i32_i32(const int32_t * x, int * dst, const int ncols, const int ncols_pad) {
-    extern __shared__ int shared_mem[];
-    int * indices = shared_mem;
-
-    const int tid = threadIdx.x;
-    const int row = blockIdx.y;
-
-    // Initialize all indices, handling the case where threads < ncols_pad
-    for (int i = tid; i < ncols_pad; i += blockDim.x) {
-        indices[i] = i < ncols ? i : 0; // Use 0 for padding indices
-    }
-    __syncthreads();
-
-    // Bitonic sort
-    for (int k = 2; k <= ncols_pad; k *= 2) {
-        for (int j = k/2; j > 0; j /= 2) {
-            for (int i = tid; i < ncols_pad; i += blockDim.x) {
-                const int ij = i ^ j;
-                if (ij > i) {
-                    // Only compare values within the actual data range
-                    if (i < ncols && ij < ncols) {
-                        if ((i & k) == 0) {
-                            if (order == GGML_SORT_ORDER_ASC) {
-                                if (x[row * ncols + indices[i]] > x[row * ncols + indices[ij]]) {
-                                    int tmp = indices[i];
-                                    indices[i] = indices[ij];
-                                    indices[ij] = tmp;
-                                }
-                            } else {
-                                if (x[row * ncols + indices[i]] < x[row * ncols + indices[ij]]) {
-                                    int tmp = indices[i];
-                                    indices[i] = indices[ij];
-                                    indices[ij] = tmp;
-                                }
-                            }
-                        } else {
-                            if (order == GGML_SORT_ORDER_ASC) {
-                                if (x[row * ncols + indices[i]] < x[row * ncols + indices[ij]]) {
-                                    int tmp = indices[i];
-                                    indices[i] = indices[ij];
-                                    indices[ij] = tmp;
-                                }
-                            } else {
-                                if (x[row * ncols + indices[i]] > x[row * ncols + indices[ij]]) {
-                                    int tmp = indices[i];
-                                    indices[i] = indices[ij];
-                                    indices[ij] = tmp;
-                                }
-                            }
-                        }
-                    }
-                }
-            }
-            __syncthreads();
-        }
-    }
-
-    // Write sorted indices to output, only threads handling valid data
-    for (int i = tid; i < ncols; i += blockDim.x) {
-        dst[row * ncols + i] = indices[i];
-    }
-}
-
-static void argsort_i32_i32_cuda(const int32_t * x, int * dst, const int ncols, const int nrows, ggml_sort_order order, cudaStream_t stream) {
-    // Bitonic sort requires ncols to be power of 2
-    const int ncols_pad = next_power_of_2(ncols);
-
-    // Ensure thread count doesn't exceed maximum (typically 1024)
-    const int max_threads = 1024;  // This is the typical max for most GPUs
-    const int threads_per_block = ncols_pad > max_threads ? max_threads : ncols_pad;
-
-    const dim3 block_dims(threads_per_block, 1, 1);
-    const dim3 block_nums(1, nrows, 1);
-    const size_t shared_mem = ncols_pad * sizeof(int);
-
-    // Check if shared memory size is within limits
-    const size_t max_shared_mem = ggml_cuda_info().devices[ggml_cuda_get_device()].smpb;
-
-    // Instead of logging an error, use GGML_ASSERT with a descriptive message
-    GGML_ASSERT(shared_mem <= max_shared_mem && "argsort: required shared memory exceeds device limit");
-
-    // Launch kernels with the updated thread configuration
-    if (order == GGML_SORT_ORDER_ASC) {
-        k_argsort_i32_i32<GGML_SORT_ORDER_ASC><<<block_nums, block_dims, shared_mem, stream>>>(x, dst, ncols, ncols_pad);
-    } else if (order == GGML_SORT_ORDER_DESC) {
-        k_argsort_i32_i32<GGML_SORT_ORDER_DESC><<<block_nums, block_dims, shared_mem, stream>>>(x, dst, ncols, ncols_pad);
-    } else {
-        GGML_ABORT("fatal error");
-    }
-}
-
-
 void ggml_cuda_op_argsort(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
     const ggml_tensor * src0 = dst->src[0];
     const float * src0_d = (const float *)src0->data;
     float * dst_d = (float *)dst->data;
     cudaStream_t stream = ctx.stream();
 
-    GGML_ASSERT(src0->type == GGML_TYPE_F32 || src0->type == GGML_TYPE_I32);
+    GGML_ASSERT(src0->type == GGML_TYPE_F32);
     GGML_ASSERT( dst->type == GGML_TYPE_I32);
     GGML_ASSERT(ggml_is_contiguous(src0));
 
@@ -194,9 +100,5 @@ void ggml_cuda_op_argsort(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
 
     enum ggml_sort_order order = (enum ggml_sort_order) dst->op_params[0];
 
-    if (src0->type == GGML_TYPE_I32) {
-        argsort_i32_i32_cuda((const int32_t *)src0_d, (int *)dst_d, ncols, nrows, order, stream);
-    } else {
-        argsort_f32_i32_cuda(src0_d, (int *)dst_d, ncols, nrows, order, stream);
-    }
+    argsort_f32_i32_cuda(src0_d, (int *)dst_d, ncols, nrows, order, stream);
 }

ml/backend/ggml/ggml/src/ggml-cuda/cpy.cu

@@ -38,13 +38,6 @@ static __device__ void cpy_1_f16_f32(const char * cxi, char * cdsti) {
     *dsti = *xi;
 }
 
-static __device__ void cpy_1_i32_i32(const char * cxi, char * cdsti) {
-    const int32_t * xi = (const int32_t *) cxi;
-    int32_t * dsti = (int32_t *) cdsti;
-
-    *dsti = *xi;
-}
-
 template <cpy_kernel_t cpy_1>
 static __global__ void cpy_f32_f16(const char * cx, char * cdst_direct, const int ne,
                                    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
@@ -75,44 +68,6 @@ static __global__ void cpy_f32_f16(const char * cx, char * cdst_direct, const in
     cpy_1(cx + x_offset, cdst + dst_offset);
 }
 
-// First, add this template function after the other template functions
-template <cpy_kernel_t cpy_1>
-static __global__ void cpy_i32_i32(const char * cx, char * cdst, const int ne,
-                                 const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
-                                 const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11,
-                                 const int nb12, const int nb13) {
-    const int64_t i = blockDim.x*blockIdx.x + threadIdx.x;
-
-    if (i >= ne) {
-        return;
-    }
-
-    const int64_t i03 = i/(ne00 * ne01 * ne02);
-    const int64_t i02 = (i - i03*ne00*ne01*ne02 )/ (ne00*ne01);
-    const int64_t i01 = (i - i03*ne00*ne01*ne02  -  i02*ne01*ne00) / ne00;
-    const int64_t i00 = i - i03*ne00*ne01*ne02 - i02*ne01*ne00 - i01*ne00;
-    const int64_t x_offset = i00*nb00 + i01*nb01 + i02*nb02 + i03 * nb03;
-
-    const int64_t i13 = i/(ne10 * ne11 * ne12);
-    const int64_t i12 = (i - i13*ne10*ne11*ne12) / (ne10*ne11);
-    const int64_t i11 = (i - i13*ne10*ne11*ne12 - i12*ne10*ne11) / ne10;
-    const int64_t i10 = i - i13*ne10*ne11*ne12 - i12*ne10*ne11 - i11*ne10;
-    const int64_t dst_offset = i10*nb10 + i11*nb11 + i12*nb12 + i13 * nb13;
-
-    cpy_1(cx + x_offset, cdst + dst_offset);
-}
-
-// Then modify the ggml_cpy_i32_i32_cuda function to use the new template
-static void ggml_cpy_i32_i32_cuda(
-    const char * cx, char * cdst, const int ne,
-    const int ne00, const int ne01, const int ne02, const int nb00, const int nb01, const int nb02,
-    const int nb03, const int ne10, const int ne11, const int ne12, const int nb10, const int nb11, const int nb12, const int nb13, cudaStream_t stream, char ** cdst_indirect, int graph_cpynode_index) {
-
-    const int num_blocks = (ne + CUDA_CPY_BLOCK_SIZE - 1) / CUDA_CPY_BLOCK_SIZE;
-    cpy_i32_i32<cpy_1_i32_i32><<<num_blocks, CUDA_CPY_BLOCK_SIZE, 0, stream>>>
-        (cx, cdst, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13);
-}
-
 static __device__ void cpy_blck_f32_q8_0(const char * cxi, char * cdsti) {
     const float * xi = (const float *) cxi;
     block_q8_0 * dsti = (block_q8_0 *) cdsti;
@@ -678,8 +633,6 @@ void ggml_cuda_cpy(ggml_backend_cuda_context & ctx, const ggml_tensor * src0, gg
         ggml_cpy_f16_f16_cuda (src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream, dest_ptrs_d, graph_cpynode_index);
     } else if (src0->type == GGML_TYPE_F16 && src1->type == GGML_TYPE_F32) {
         ggml_cpy_f16_f32_cuda (src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream, dest_ptrs_d, graph_cpynode_index);
-    } else if (src0->type == GGML_TYPE_I32 && src1->type == GGML_TYPE_I32) {
-        ggml_cpy_i32_i32_cuda(src0_ddc, src1_ddc, ne, ne00, ne01, ne02, nb00, nb01, nb02, nb03, ne10, ne11, ne12, nb10, nb11, nb12, nb13, main_stream, dest_ptrs_d, graph_cpynode_index);
     } else {
         GGML_ABORT("%s: unsupported type combination (%s to %s)\n", __func__,
                 ggml_type_name(src0->type), ggml_type_name(src1->type));
@@ -735,8 +688,6 @@ void* ggml_cuda_cpy_fn(const ggml_tensor * src0, ggml_tensor * src1) {
         return (void*) cpy_f32_f16<cpy_1_f32_f16>;
     } else if (src0->type == GGML_TYPE_F16 && src1->type == GGML_TYPE_F32) {
         return (void*) cpy_f32_f16<cpy_1_f16_f32>;
-    } else if (src0->type == GGML_TYPE_I32 && src1->type == GGML_TYPE_I32) {
-        return (void*) cpy_i32_i32<cpy_1_i32_i32>;
     } else {
         GGML_ABORT("%s: unsupported type combination (%s to %s)\n", __func__,
                 ggml_type_name(src0->type), ggml_type_name(src1->type));

model/input/input.go

@@ -2,30 +2,16 @@ package input
 
 import "github.com/ollama/ollama/ml"
 
-// Multimodal is a multimodal embedding or a component of one.
-// For example, it could be a row of an image that can be processed
-// independently.
-type Multimodal struct {
-	// Tensor is the embedding data. Implementations may chose what to
-	// store here or it may be nil if not needed. However, any ml.Tensor
-	// objects must be stored here and not in Data.
-	Tensor ml.Tensor
-
-	// Data is implementation-specific opaque data, such as metadata on how
-	// to layout Tensor. It may be nil if not needed. It may also store larger
-	// objects such as complete images if they are to be processed later.
-	Data any
-}
-
 // Input represents one token in the input stream
 type Input struct {
 	// Token is a single element of text.
 	Token int32
 
-	// Multimodal is represents a non-text element such as an
-	// image (or part of one if the image can be processed in pieces).
-	// It may be used either together with Token or on its own.
-	Multimodal []Multimodal
+	// Multimodal is opaque data representing a non-text
+	// element such as an image (or part of one if the image
+	// can be processed in pieces). It may be either together
+	// with Token or on its own.
+	Multimodal any
 
 	// MultimodalHash is a unique representation of the data
 	// stored in Multimodal, used for caching and comparing
@@ -46,7 +32,7 @@ type Input struct {
 // Positions slice.
 type MultimodalIndex struct {
 	Index      int
-	Multimodal []Multimodal
+	Multimodal any
 }
 
 // Batch contains the inputs for a model forward pass

model/model.go

@@ -40,13 +40,12 @@ type MultimodalProcessor interface {
 	// EncodeMultimodal processes a single input (such as an image) and
 	// generates an output (typically an embedding) that can be used by the model.
 	//
-	// The return value is one or more tensors, each with optional model-specific
-	// opaque metadata. Typically, the tensors might be views into an embedding
-	// with each view representing a chunk of data that can be processed independently
-	// in different batches.
+	// The return value is most typically an ml.Tensor, however, different
+	// type are possible, such as an object containing a tensor plus
+	// additional metadata, a slice of tensors or even just the original input.
 	//
 	// The result may be cached by the runner.
-	EncodeMultimodal(ml.Context, []byte) ([]input.Multimodal, error)
+	EncodeMultimodal(ml.Context, []byte) (any, error)
 
 	// PostTokenize is called after tokenization to allow the model to edit the
 	// input stream to correctly arrange multimodal elements.

model/models/gemma3/model.go

@@ -82,7 +82,7 @@ func New(c fs.Config) (model.Model, error) {
 	return &m, nil
 }
 
-func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input.Multimodal, error) {
+func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) (any, error) {
 	if len(m.VisionModel.Layers) == 0 {
 		return nil, model.ErrNoVisionModel
 	}
@@ -108,22 +108,22 @@ func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input
 
 	visionOutputs := m.VisionModel.Forward(ctx, pixelValues)
 	visionOutputs = m.MultiModalProjector.Forward(ctx, visionOutputs, m.imageSize, m.patchSize, m.VisionModel.eps)
-	return []input.Multimodal{{Tensor: visionOutputs}}, nil
+	return visionOutputs, nil
 }
 
 func (m *Model) PostTokenize(inputs []input.Input) ([]input.Input, error) {
 	var result []input.Input
 
 	for _, inp := range inputs {
-		if len(inp.Multimodal) == 0 {
+		if inp.Multimodal == nil {
 			result = append(result, inp)
 		} else {
-			inputMultimodal := inp.Multimodal[0].Tensor
+			inputMultimodal := inp.Multimodal.(ml.Tensor)
 
 			result = append(result,
-				input.Input{Token: 108, SameBatch: inputMultimodal.Dim(1) + 3}, // "\n\n"
-				input.Input{Token: 255999},                                     // "<start_of_image>""
-				input.Input{Multimodal: []input.Multimodal{{Tensor: inputMultimodal}}, MultimodalHash: inp.MultimodalHash}, // image data is on the first placeholder
+				input.Input{Token: 108, SameBatch: inputMultimodal.Dim(1) + 3},               // "\n\n"
+				input.Input{Token: 255999},                                                   // "<start_of_image>""
+				input.Input{Multimodal: inputMultimodal, MultimodalHash: inp.MultimodalHash}, // image data is on the first placeholder
 			)
 
 			// add image token placeholders

model/models/gemma3/model_text.go

@@ -165,7 +165,7 @@ func (m *TextModel) Forward(ctx ml.Context, inputs, positions, outputs ml.Tensor
 	// set image embeddings
 	var except []int
 	for _, image := range batch.Multimodal {
-		visionOutputs := image.Multimodal[0].Tensor
+		visionOutputs := image.Multimodal.(ml.Tensor)
 		ctx.Forward(visionOutputs.Copy(ctx, hiddenState.View(ctx, image.Index*hiddenState.Stride(1), visionOutputs.Dim(0)*visionOutputs.Dim(1))))
 
 		for i := range visionOutputs.Dim(1) {

model/models/llama4/model.go

@@ -4,6 +4,7 @@ import (
 	"bytes"
 	"image"
 	"slices"
+	"sync"
 
 	"github.com/ollama/ollama/fs"
 	"github.com/ollama/ollama/kvcache"
@@ -62,7 +63,7 @@ func New(c fs.Config) (model.Model, error) {
 	return &m, nil
 }
 
-func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input.Multimodal, error) {
+func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) (any, error) {
 	if len(m.VisionModel.Layers) < 1 {
 		return nil, model.ErrNoVisionModel
 	}
@@ -102,79 +103,70 @@ func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input
 	visionOutputs := m.VisionModel.Forward(ctx, pixelValues)
 	visionOutputs = visionOutputs.Reshape(ctx, visionOutputs.Dim(0), visionOutputs.Dim(1)*visionOutputs.Dim(2)*visionOutputs.Dim(3))
 	projectedOutputs := m.Projector.Forward(ctx, visionOutputs)
-
-	var multimodal []input.Multimodal
-	aspectRatio := image.Point{ratioW, ratioH}
-
-	var offset int
-	patchesPerChunk := projectedOutputs.Dim(1)
-	if aspectRatio.Y*aspectRatio.X > 1 {
-		patchesPerChunk = projectedOutputs.Dim(1) / (aspectRatio.X*aspectRatio.Y + 1)
-
-		for range aspectRatio.Y {
-			for x := range aspectRatio.X {
-				view := projectedOutputs.View(ctx, projectedOutputs.Stride(1)*offset,
-					projectedOutputs.Dim(0), projectedOutputs.Stride(1),
-					patchesPerChunk)
-				var separator separator
-				if x < aspectRatio.X-1 {
-					separator.x = true // <|tile_x_separator|>
-				} else {
-					separator.y = true // <|tile_y_separator|>
-				}
-				multimodal = append(multimodal, input.Multimodal{Tensor: view, Data: &separator})
-				offset += patchesPerChunk
-			}
-		}
-	}
-
-	view := projectedOutputs.View(ctx, projectedOutputs.Stride(1)*offset,
-		projectedOutputs.Dim(0), projectedOutputs.Stride(1),
-		patchesPerChunk)
-	multimodal = append(multimodal, input.Multimodal{Tensor: view, Data: &separator{}})
-
-	return multimodal, nil
+	return &chunks{Model: m, Tensor: projectedOutputs, aspectRatio: image.Point{ratioW, ratioH}}, nil
 }
 
-type separator struct {
-	x bool
-	y bool
+type chunks struct {
+	*Model
+	ml.Tensor
+	aspectRatio image.Point
+
+	dataOnce sync.Once
+	data     []float32
+}
+
+type chunk struct {
+	*chunks
+	s, n int
+}
+
+func (r *chunk) floats() []float32 {
+	r.dataOnce.Do(func() {
+		temp := r.Backend().NewContext()
+		defer temp.Close()
+		temp.Forward(r.Tensor).Compute(r.Tensor)
+		r.data = r.Floats()
+	})
+
+	return r.data[r.s*r.Dim(0) : (r.s+r.n)*r.Dim(0)]
 }
 
 func (m *Model) PostTokenize(inputs []input.Input) ([]input.Input, error) {
 	var result []input.Input
 	for _, inp := range inputs {
-		if len(inp.Multimodal) == 0 {
+		if inp.Multimodal == nil {
 			result = append(result, inp)
 			continue
 		}
 
+		t := inp.Multimodal.(*chunks)
 		var imageInputs []input.Input
 		imageInputs = append(imageInputs, input.Input{Token: 200080}) // <|image_start|>
 
-		for i, mm := range inp.Multimodal {
-			patchesPerChunk := mm.Tensor.Dim(1)
+		var offset int
+		patchesPerChunk := t.Dim(1)
+		if t.aspectRatio.Y*t.aspectRatio.X > 1 {
+			patchesPerChunk = t.Dim(1) / (t.aspectRatio.X*t.aspectRatio.Y + 1)
 
-			if i < len(inp.Multimodal)-1 {
-				separator := mm.Data.(*separator)
-
-				imageInputs = append(imageInputs, input.Input{Token: 200092, Multimodal: []input.Multimodal{{Tensor: mm.Tensor}}, MultimodalHash: inp.MultimodalHash, SameBatch: patchesPerChunk}) // <|patch|>
-				imageInputs = append(imageInputs, slices.Repeat([]input.Input{{Token: 200092}}, patchesPerChunk-1)...)
-
-				if separator.x {
-					imageInputs = append(imageInputs, input.Input{Token: 200084}) // <|tile_x_separator|>
+			for range t.aspectRatio.Y {
+				for x := range t.aspectRatio.X {
+					imageInputs = append(imageInputs, input.Input{Token: 200092, Multimodal: &chunk{t, offset, patchesPerChunk}, MultimodalHash: inp.MultimodalHash, SameBatch: patchesPerChunk}) // <|patch|>
+					imageInputs = append(imageInputs, slices.Repeat([]input.Input{{Token: 200092}}, patchesPerChunk-1)...)
+					if x < t.aspectRatio.X-1 {
+						imageInputs = append(imageInputs, input.Input{Token: 200084}) // <|tile_x_separator|>
+					}
+					offset += patchesPerChunk
 				}
-				if separator.y {
-					imageInputs = append(imageInputs, input.Input{Token: 200085}) // <|tile_y_separator|>
-				}
-			} else {
-				imageInputs = append(imageInputs, input.Input{Token: 200090})                                                                                                                      // <|image|>
-				imageInputs = append(imageInputs, input.Input{Token: 200092, Multimodal: []input.Multimodal{{Tensor: mm.Tensor}}, MultimodalHash: inp.MultimodalHash, SameBatch: patchesPerChunk}) // <|patch|>
-				imageInputs = append(imageInputs, slices.Repeat([]input.Input{{Token: 200092}}, patchesPerChunk-1)...)
-				imageInputs = append(imageInputs, input.Input{Token: 200080}) // <|image_end|>
+
+				imageInputs = append(imageInputs, input.Input{Token: 200085}) // <|tile_y_separator|>
 			}
 		}
 
+		imageInputs = append(imageInputs, input.Input{Token: 200090})                                                                                                                 // <|image|>
+		imageInputs = append(imageInputs, input.Input{Token: 200092, Multimodal: &chunk{t, offset, patchesPerChunk}, MultimodalHash: inp.MultimodalHash, SameBatch: patchesPerChunk}) // <|patch|>
+		imageInputs = append(imageInputs, slices.Repeat([]input.Input{{Token: 200092}}, patchesPerChunk-1)...)
+		imageInputs = append(imageInputs, input.Input{Token: 200080}) // <|image_end|>
+
 		result = append(result, imageInputs...)
 	}
 

model/models/llama4/model_text.go

@@ -210,7 +210,12 @@ func (m *TextModel) Forward(ctx ml.Context, inputs, positions, outputs ml.Tensor
 	hiddenStates := m.TokenEmbedding.Forward(ctx, inputs).Duplicate(ctx)
 
 	for _, mi := range batch.Multimodal {
-		img := mi.Multimodal[0].Tensor
+		f32s := mi.Multimodal.(*chunk).floats()
+		img, err := ctx.Input().FromFloatSlice(f32s, len(f32s)/m.hiddenSize, m.hiddenSize)
+		if err != nil {
+			panic(err)
+		}
+
 		ctx.Forward(img.Copy(ctx, hiddenStates.View(ctx, mi.Index*hiddenStates.Stride(1), img.Dim(0)*img.Dim(1))))
 	}
 

model/models/mistral3/model.go

@@ -4,6 +4,7 @@ import (
 	"bytes"
 	"image"
 	"slices"
+	"sync"
 
 	"github.com/ollama/ollama/fs"
 	"github.com/ollama/ollama/kvcache"
@@ -104,7 +105,7 @@ func newMultiModalProjector(c fs.Config) *MultiModalProjector {
 	}
 }
 
-func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input.Multimodal, error) {
+func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) (any, error) {
 	if len(m.VisionModel.Layers) == 0 {
 		return nil, model.ErrNoVisionModel
 	}
@@ -128,14 +129,37 @@ func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input
 	features, size := m.MultiModalProjector.Forward(ctx, visionOutputs, size)
 
 	// split into patches to be sent to the text transformer
-	rows := make([]input.Multimodal, size.Y)
+	parent := imageFeatures{tensor: features}
+	rows := make([]*imageRow, size.Y)
 	for i := range rows {
-		rows[i].Tensor = features.View(ctx, features.Stride(1)*size.X*i, features.Dim(0), features.Stride(1), size.X)
+		rows[i] = &imageRow{parent: &parent, s: i, shape: []int{features.Dim(0), size.X}}
 	}
 
 	return rows, nil
 }
 
+type imageFeatures struct {
+	tensor ml.Tensor
+
+	dataOnce sync.Once
+	data     []float32
+}
+
+type imageRow struct {
+	parent *imageFeatures
+	s      int
+	shape  []int
+}
+
+func (r *imageRow) data() []float32 {
+	n := 1
+	for _, s := range r.shape {
+		n *= s
+	}
+
+	return r.parent.data[r.s*n : (r.s+1)*n]
+}
+
 // PostTokenize arranges Mistral 3's inputs for the forward pass
 // In Mistral 3 and Pixtral, the input patches are arranged as follows:
 // [IMG]...[IMG][IMG_BREAK][IMG]...[IMG][IMG_BREAK][IMG]...[IMG][IMG_END]
@@ -144,14 +168,15 @@ func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input
 func (m *Model) PostTokenize(inputs []input.Input) ([]input.Input, error) {
 	var result []input.Input
 	for _, inp := range inputs {
-		if len(inp.Multimodal) == 0 {
+		if inp.Multimodal == nil {
 			result = append(result, inp)
 		} else {
-			for i, row := range inp.Multimodal {
+			inputMultimodal := inp.Multimodal.([]*imageRow)
+			for i, row := range inputMultimodal {
 				// [IMG]
-				result = append(result, input.Input{Token: 10, Multimodal: []input.Multimodal{{Tensor: row.Tensor}}, MultimodalHash: inp.MultimodalHash, SameBatch: row.Tensor.Dim(1)})
-				result = append(result, slices.Repeat([]input.Input{{Token: 10}}, row.Tensor.Dim(1)-1)...)
-				if i == len(inp.Multimodal)-1 {
+				result = append(result, input.Input{Token: 10, Multimodal: row, MultimodalHash: inp.MultimodalHash, SameBatch: row.shape[1]})
+				result = append(result, slices.Repeat([]input.Input{{Token: 10}}, row.shape[1]-1)...)
+				if i == len(inputMultimodal)-1 {
 					// [IMG_END]
 					result = append(result, input.Input{Token: 13})
 				} else {

model/models/mistral3/model_text.go

@@ -9,6 +9,7 @@ import (
 	"github.com/ollama/ollama/kvcache"
 	"github.com/ollama/ollama/ml"
 	"github.com/ollama/ollama/ml/nn"
+	"github.com/ollama/ollama/model"
 	"github.com/ollama/ollama/model/input"
 )
 
@@ -19,6 +20,8 @@ type TextOptions struct {
 }
 
 type TextModel struct {
+	model.Base
+
 	TokenEmbedding *nn.Embedding `gguf:"token_embd"`
 	Layers         []Layer       `gguf:"blk"`
 	OutputNorm     *nn.RMSNorm   `gguf:"output_norm"`
@@ -106,7 +109,20 @@ func (m *TextModel) Forward(ctx ml.Context, inputs, positions, outputs ml.Tensor
 
 	// image embeddings
 	for _, image := range batch.Multimodal {
-		imageFeature := image.Multimodal[0].Tensor
+		row := image.Multimodal.(*imageRow)
+		row.parent.dataOnce.Do(func() {
+			// use a new, throwaway context so the image tensor is not added to the graph
+			temp := m.Backend().NewContext()
+			temp.Forward(row.parent.tensor).Compute(row.parent.tensor)
+			row.parent.data = row.parent.tensor.Floats()
+			temp.Close()
+		})
+
+		imageFeature, err := ctx.Input().FromFloatSlice(row.data(), row.shape...)
+		if err != nil {
+			panic(err)
+		}
+
 		ctx.Forward(imageFeature.Copy(ctx, hiddenState.View(ctx, image.Index*hiddenState.Stride(1), imageFeature.Dim(0)*imageFeature.Dim(1))))
 	}
 

model/models/mllama/model.go

@@ -3,7 +3,6 @@ package mllama
 import (
 	"bytes"
 	"image"
-	"slices"
 
 	"github.com/ollama/ollama/fs"
 	"github.com/ollama/ollama/kvcache"
@@ -59,7 +58,7 @@ func New(c fs.Config) (model.Model, error) {
 	return &m, nil
 }
 
-func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input.Multimodal, error) {
+func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) (any, error) {
 	if len(m.VisionModel.Transformer.Layers) == 0 || len(m.GlobalTransformer.Layers) == 0 {
 		return nil, model.ErrNoVisionModel
 	}
@@ -74,17 +73,13 @@ func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input
 		return nil, err
 	}
 
-	if ratio.numTiles() < m.maxNumTiles {
-		// Pad tiles to maxNumTiles
-		f32s = slices.Grow(f32s, m.imageSize*m.imageSize*m.numChannels*m.maxNumTiles)
-		f32s = f32s[:m.imageSize*m.imageSize*m.numChannels*m.maxNumTiles]
-	}
-
-	pixelValues, err := ctx.Input().FromFloatSlice(f32s, m.imageSize, m.imageSize, m.numChannels, m.maxNumTiles)
+	pixelValues, err := ctx.Input().FromFloatSlice(f32s, m.imageSize, m.imageSize, m.numChannels, ratio.numTiles())
 	if err != nil {
 		return nil, err
 	}
 
+	pixelValues = pixelValues.Pad(ctx, 0, 0, 0, m.ImageProcessor.maxNumTiles-ratio.numTiles())
+
 	aspectRatio, err := ctx.Input().FromIntSlice([]int32{int32(ratio.rank)}, 1)
 	if err != nil {
 		return nil, err
@@ -92,9 +87,7 @@ func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input
 
 	positionIDs := ctx.Arange(0, 1601, 1, ml.DTypeI32)
 	crossAttentionStates := m.VisionModel.Forward(ctx, pixelValues, positionIDs, aspectRatio)
-	projectedOutputs := m.Projector.Forward(ctx, crossAttentionStates)
-
-	return []input.Multimodal{{Tensor: projectedOutputs}}, nil
+	return m.Projector.Forward(ctx, crossAttentionStates), nil
 }
 
 func (m *Model) PostTokenize(inputs []input.Input) ([]input.Input, error) {
@@ -110,7 +103,7 @@ func (m *Model) PostTokenize(inputs []input.Input) ([]input.Input, error) {
 func (m *Model) Forward(ctx ml.Context, batch input.Batch) (ml.Tensor, error) {
 	var crossAttentionStates ml.Tensor
 	if len(batch.Multimodal) > 0 {
-		crossAttentionStates = batch.Multimodal[len(batch.Multimodal)-1].Multimodal[0].Tensor
+		crossAttentionStates = batch.Multimodal[len(batch.Multimodal)-1].Multimodal.(ml.Tensor)
 	}
 
 	positions, err := ctx.Input().FromIntSlice(batch.Positions, len(batch.Positions))

model/models/models.go

@@ -7,5 +7,4 @@ import (
 	_ "github.com/ollama/ollama/model/models/llama4"
 	_ "github.com/ollama/ollama/model/models/mistral3"
 	_ "github.com/ollama/ollama/model/models/mllama"
-	_ "github.com/ollama/ollama/model/models/qwen25vl"
 )

model/models/qwen25vl/model.go

@@ -1,158 +0,0 @@
-package qwen25vl
-
-import (
-	"bytes"
-	"fmt"
-	"image"
-	"slices"
-
-	"github.com/ollama/ollama/fs"
-	"github.com/ollama/ollama/kvcache"
-	"github.com/ollama/ollama/ml"
-	"github.com/ollama/ollama/model"
-	"github.com/ollama/ollama/model/input"
-)
-
-type Model struct {
-	model.Base
-	model.BytePairEncoding
-
-	*TextModel
-	*VisionModel `gguf:"v,vision"`
-
-	ImageProcessor
-}
-
-// Implement MultimodalProcessor interface
-var _ model.MultimodalProcessor = (*Model)(nil)
-
-func New(c fs.Config) (model.Model, error) {
-	m := &Model{
-		BytePairEncoding: model.NewBytePairEncoding(
-			c.String("tokenizer.ggml.pretokenizer", `(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\r\n\p{L}\p{N}]?\p{L}+|\p{N}| ?[^\s\p{L}\p{N}]+[\r\n]*|\s*[\r\n]+|\s+(?!\S)|\s+`),
-			&model.Vocabulary{
-				Values: c.Strings("tokenizer.ggml.tokens"),
-				Types:  c.Ints("tokenizer.ggml.token_type"),
-				Merges: c.Strings("tokenizer.ggml.merges"),
-				BOS:    int32(c.Uint("tokenizer.ggml.bos_token_id")),
-				AddBOS: c.Bool("tokenizer.ggml.add_bos_token", false),
-				EOS:    int32(c.Uint("tokenizer.ggml.eos_token_id")),
-				AddEOS: c.Bool("tokenizer.ggml.add_eos_token", false),
-				EOT:    int32(c.Uint("tokenizer.ggml.eos_token_id")),
-				AddEOT: c.Bool("tokenizer.ggml.add_eos_token", false),
-			},
-		),
-		TextModel:      NewTextModel(c),
-		VisionModel:    newVisionModel(c),
-		ImageProcessor: newImageProcessor(c),
-	}
-
-	m.Cache = kvcache.NewCausalCache(m.TextModel.Shift)
-
-	return m, nil
-}
-
-func (m *Model) PixelValues(ctx ml.Context, multimodalData []byte) (ml.Tensor, *Grid, error) {
-	image, _, err := image.Decode(bytes.NewReader(multimodalData))
-	if err != nil {
-		return nil, nil, err
-	}
-
-	f32s, grid, err := m.ImageProcessor.ProcessImage(image)
-	if err != nil {
-		return nil, nil, err
-	}
-
-	// Calculate tensor dimensions
-	patchDim := m.ImageProcessor.numChannels * m.ImageProcessor.temporalPatchSize *
-		m.ImageProcessor.patchSize * m.ImageProcessor.patchSize
-	numPatches := grid.Temporal * grid.Height * grid.Width
-
-	pixelValues, err := ctx.Input().FromFloatSlice(f32s, patchDim, numPatches)
-	if err != nil {
-		return nil, nil, fmt.Errorf("failed to create tensor from image: %w", err)
-	}
-
-	return pixelValues, grid, nil
-}
-
-func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input.Multimodal, error) {
-	if len(m.VisionModel.Layers) == 0 {
-		return nil, model.ErrNoVisionModel
-	}
-
-	pixels, grid, err := m.PixelValues(ctx, multimodalData)
-	if err != nil {
-		return nil, err
-	}
-
-	visionOutputs := m.VisionModel.Forward(ctx, pixels, grid)
-	return []input.Multimodal{{Tensor: visionOutputs}}, nil
-}
-
-// PostTokenize arranges Qwen-2.5-VL's inputs for the forward pass
-func (m *Model) PostTokenize(inputs []input.Input) ([]input.Input, error) {
-	var result []input.Input
-
-	var (
-		imageToken       int32 = 151655
-		visionStartToken int32 = 151652
-		visionEndToken   int32 = 151653
-	)
-
-	nImg := 0
-	for _, inp := range inputs {
-		if inp.Multimodal == nil {
-			// If not a multimodal input, add it to the result unchanged
-			result = append(result, inp)
-		} else {
-			// Adding the 'Picture' prefix is a hack, at the time of writing there is no way to prefix
-			// the image tokens with a prompt, so we add a prefix here
-			nImg++
-			pre, err := m.Encode(fmt.Sprintf(" Picture %d: ", nImg), true)
-			if err != nil {
-				return nil, fmt.Errorf("failed to encode image prompt: %w", err)
-			}
-			for i := range pre {
-				result = append(result, input.Input{Token: pre[i]})
-			}
-
-			patchesPerChunk := inp.Multimodal[0].Tensor.Dim(1)
-
-			// First add the vision start token
-			result = append(result, input.Input{Token: visionStartToken, SameBatch: patchesPerChunk + 1})
-
-			// Add the image token with the multimodal tensor data at the first position
-			result = append(result, input.Input{
-				Token:          imageToken,
-				Multimodal:     inp.Multimodal,
-				MultimodalHash: inp.MultimodalHash,
-			})
-
-			// Add the placeholder tokens for the remaining positions (tokensPerGrid-1)
-			result = append(result, slices.Repeat([]input.Input{{Token: imageToken}}, patchesPerChunk-1)...)
-
-			result = append(result, input.Input{Token: visionEndToken})
-		}
-	}
-
-	return result, nil
-}
-
-func (m *Model) Forward(ctx ml.Context, batch input.Batch) (ml.Tensor, error) {
-	positions, err := ctx.Input().FromIntSlice(batch.Positions, len(batch.Positions))
-	if err != nil {
-		return nil, err
-	}
-
-	outputs, err := ctx.Input().FromIntSlice(batch.Outputs, len(batch.Outputs))
-	if err != nil {
-		return nil, err
-	}
-
-	return m.TextModel.Forward(ctx, batch.Inputs, positions, outputs, batch, m.Cache)
-}
-
-func init() {
-	model.Register("qwen25vl", New)
-}

model/models/qwen25vl/model_text.go

@@ -1,150 +0,0 @@
-package qwen25vl
-
-import (
-	"math"
-
-	"github.com/ollama/ollama/fs"
-	"github.com/ollama/ollama/kvcache"
-	"github.com/ollama/ollama/ml"
-	"github.com/ollama/ollama/ml/nn"
-	"github.com/ollama/ollama/model/input"
-)
-
-type TextOptions struct {
-	ctxLen, hiddenSize, numHeads, numKVHeads int
-	eps, ropeBase, ropeScale                 float32
-	ropeDim, defaultContextLen               uint32
-}
-
-type TextModel struct {
-	TokenEmbedding *nn.Embedding `gguf:"token_embd"`
-	Layers         []Layer       `gguf:"blk"`
-	OutputNorm     *nn.RMSNorm   `gguf:"output_norm"`
-	Output         *nn.Linear    `gguf:"output,alt:token_embd"`
-
-	*TextOptions
-}
-
-func NewTextModel(c fs.Config) *TextModel {
-	m := TextModel{
-		Layers: make([]Layer, c.Uint("block_count")),
-		TextOptions: &TextOptions{
-			ctxLen:            int(c.Uint("context_length")),
-			hiddenSize:        int(c.Uint("embedding_length")),
-			numHeads:          int(c.Uint("attention.head_count")),
-			numKVHeads:        int(c.Uint("attention.head_count_kv")),
-			eps:               c.Float("attention.layer_norm_rms_epsilon"),
-			ropeBase:          c.Float("rope.freq_base"),
-			ropeScale:         c.Float("rope.freq_scale", 1),
-			ropeDim:           c.Uint("rope.dimension_count", 128),
-			defaultContextLen: c.Uint("context_length", 128000),
-		},
-	}
-
-	return &m
-}
-
-// SelfAttention implements the multi-head self-attention mechanism
-// with separate projections for query, key, value and output transformations
-type SelfAttention struct {
-	Query  *nn.Linear `gguf:"attn_q"`
-	Key    *nn.Linear `gguf:"attn_k"`
-	Value  *nn.Linear `gguf:"attn_v"`
-	Output *nn.Linear `gguf:"attn_output"`
-}
-
-func (sa *SelfAttention) Forward(ctx ml.Context, hiddenState, positionIDs ml.Tensor, cache kvcache.Cache, opts *TextOptions) ml.Tensor {
-	batchSize := hiddenState.Dim(1)
-	headDim := opts.hiddenSize / opts.numHeads
-
-	q := sa.Query.Forward(ctx, hiddenState)
-	q = q.Reshape(ctx, headDim, opts.numHeads, batchSize)
-	q = q.RoPE(ctx, positionIDs, nil, opts.ropeDim, 2, opts.ropeBase, opts.ropeScale, ml.WithContextLen(opts.defaultContextLen))
-
-	k := sa.Key.Forward(ctx, hiddenState)
-	k = k.Reshape(ctx, headDim, opts.numKVHeads, batchSize)
-	k = k.RoPE(ctx, positionIDs, nil, opts.ropeDim, 2, opts.ropeBase, opts.ropeScale, ml.WithContextLen(opts.defaultContextLen))
-
-	v := sa.Value.Forward(ctx, hiddenState)
-	v = v.Reshape(ctx, headDim, opts.numKVHeads, batchSize)
-
-	scaleFactor := 1.0 / math.Sqrt(float64(headDim))
-	kqv := nn.Attention(ctx, q, k, v, scaleFactor, cache)
-	kqv = kqv.Reshape(ctx, opts.hiddenSize, batchSize)
-
-	return sa.Output.Forward(ctx, kqv)
-}
-
-// Shift applies rotary position embeddings to the key tensor for causal attention caching
-func (m *TextModel) Shift(ctx ml.Context, layer int, key, shift ml.Tensor) (ml.Tensor, error) {
-	return key.RoPE(ctx, shift, nil, m.ropeDim, 2, m.ropeBase, m.ropeScale, ml.WithContextLen(m.defaultContextLen)), nil
-}
-
-// MLP implements the feed-forward network component with SwiGLU activation
-type MLP struct {
-	Up   *nn.Linear `gguf:"ffn_up"`
-	Down *nn.Linear `gguf:"ffn_down"`
-	Gate *nn.Linear `gguf:"ffn_gate"`
-}
-
-func (mlp *MLP) Forward(ctx ml.Context, hiddenState ml.Tensor, opts *TextOptions) ml.Tensor {
-	// Apply SwiGLU activation gating
-	hiddenState = mlp.Gate.Forward(ctx, hiddenState).SILU(ctx).Mul(ctx, mlp.Up.Forward(ctx, hiddenState))
-	// Project back to hidden dimension
-	return mlp.Down.Forward(ctx, hiddenState)
-}
-
-// Layer represents a single transformer layer combining self-attention and feed-forward components
-type Layer struct {
-	AttentionNorm *nn.RMSNorm `gguf:"attn_norm"`
-	SelfAttention *SelfAttention
-	MLPNorm       *nn.RMSNorm `gguf:"ffn_norm"`
-	MLP           *MLP
-}
-
-func (l *Layer) Forward(ctx ml.Context, hiddenState, positionIDs, outputs ml.Tensor, cache kvcache.Cache, opts *TextOptions) ml.Tensor {
-	// Self-attention branch with residual connection
-	residual := hiddenState
-
-	hiddenState = l.AttentionNorm.Forward(ctx, hiddenState, opts.eps)
-	hiddenState = l.SelfAttention.Forward(ctx, hiddenState, positionIDs, cache, opts)
-
-	// In the final layer (outputs != nil), optimize by pruning to just the token positions
-	// we need logits for.
-	if outputs != nil {
-		hiddenState = hiddenState.Rows(ctx, outputs)
-		residual = residual.Rows(ctx, outputs)
-	}
-
-	hiddenState = hiddenState.Add(ctx, residual)
-	// Feed-forward branch with residual connection
-	residual = hiddenState
-	hiddenState = l.MLPNorm.Forward(ctx, hiddenState, opts.eps)
-	hiddenState = l.MLP.Forward(ctx, hiddenState, opts)
-	return hiddenState.Add(ctx, residual)
-}
-
-func (m *TextModel) Forward(ctx ml.Context, inputs, positions, outputs ml.Tensor, batch input.Batch, cache kvcache.Cache) (ml.Tensor, error) {
-	// Initial token embedding
-	hiddenStates := m.TokenEmbedding.Forward(ctx, inputs).Duplicate(ctx)
-
-	for _, mi := range batch.Multimodal {
-		img := mi.Multimodal[0].Tensor
-		ctx.Forward(img.Copy(ctx, hiddenStates.View(ctx, mi.Index*hiddenStates.Stride(1), img.Dim(0)*img.Dim(1))))
-	}
-
-	// Process through transformer layers
-	for i, layer := range m.Layers {
-		cache.SetLayer(i)
-
-		var lastLayerOutputs ml.Tensor
-		if i == len(m.Layers)-1 {
-			lastLayerOutputs = outputs
-		}
-
-		hiddenStates = layer.Forward(ctx, hiddenStates, positions, lastLayerOutputs, cache, m.TextOptions)
-	}
-
-	hiddenStates = m.OutputNorm.Forward(ctx, hiddenStates, m.eps)
-	return m.Output.Forward(ctx, hiddenStates), nil
-}

model/models/qwen25vl/model_vision.go

@@ -1,391 +0,0 @@
-package qwen25vl
-
-import (
-	"fmt"
-	"math"
-	"slices"
-
-	"github.com/ollama/ollama/fs"
-	"github.com/ollama/ollama/ml"
-	"github.com/ollama/ollama/ml/nn"
-)
-
-// We only support batch size of 1
-var batchSize int = 1
-
-func rotateHalf(ctx ml.Context, t ml.Tensor) ml.Tensor {
-	x1 := t.View(ctx, 0, t.Dim(0)/2, t.Stride(1), t.Dim(1), t.Stride(2), t.Dim(2), t.Stride(3), t.Dim(3))
-	x2 := t.View(ctx, t.Stride(0)*t.Dim(0)/2, t.Dim(0)/2, t.Stride(1), t.Dim(1), t.Stride(2), t.Dim(2), t.Stride(3), t.Dim(3)).Contiguous(ctx)
-	return x2.Neg(ctx).Concat(ctx, x1, 0)
-}
-
-func applyRotaryPositionalEmbedding(ctx ml.Context, t, cos, sin ml.Tensor) ml.Tensor {
-	return t.Mul(ctx, cos).Add(ctx, rotateHalf(ctx, t).Mul(ctx, sin))
-}
-
-func blockDiagonalMask(ctx ml.Context, seqLength int, bounds []int, numHeads int) ml.Tensor {
-	// Create a flat slice for the mask (all -inf initially to block all attention)
-	flat := make([]float32, seqLength*seqLength)
-	for i := range flat {
-		flat[i] = float32(math.Inf(-1)) // Negative infinity to block attention
-	}
-
-	// Fill in the mask with zeros for tokens that CAN attend to each other
-	for i := 1; i < len(bounds); i++ {
-		start := bounds[i-1]
-		end := bounds[i]
-
-		// Enable attention within this sequence block by setting values to 0
-		for row := start; row < end; row++ {
-			for col := start; col < end; col++ {
-				idx := row*seqLength + col
-				flat[idx] = 0.0 // 0 allows attention, -inf blocks it
-			}
-		}
-	}
-
-	mask, err := ctx.Input().FromFloatSlice(flat, seqLength, seqLength)
-	if err != nil {
-		panic(err)
-	}
-	// Reshape to match [seqLength, seqLength, 1] for broadcasting
-	mask = mask.Reshape(ctx, seqLength, seqLength, 1)
-
-	return mask
-}
-
-type VisionSelfAttention struct {
-	Query  *nn.Linear `gguf:"attn_q"`
-	Key    *nn.Linear `gguf:"attn_k"`
-	Value  *nn.Linear `gguf:"attn_v"`
-	Output *nn.Linear `gguf:"attn_out"`
-}
-
-func (sa *VisionSelfAttention) Forward(ctx ml.Context, hiddenStates, cos, sin, mask ml.Tensor, opts *VisionModelOptions) ml.Tensor {
-	query := sa.Query.Forward(ctx, hiddenStates)
-	key := sa.Key.Forward(ctx, hiddenStates)
-	value := sa.Value.Forward(ctx, hiddenStates)
-
-	query = query.Reshape(ctx, opts.headDim, opts.numHeads, query.Dim(1), batchSize)
-	key = key.Reshape(ctx, opts.headDim, opts.numHeads, key.Dim(1), batchSize)
-	value = value.Reshape(ctx, opts.headDim, opts.numHeads, value.Dim(1), batchSize)
-
-	query = applyRotaryPositionalEmbedding(ctx, query, cos, sin)
-	key = applyRotaryPositionalEmbedding(ctx, key, cos, sin)
-
-	// Scale factor for scaled dot-product attention
-	scale := 1.0 / math.Sqrt(float64(opts.headDim))
-
-	// Scaled dot-product attention
-	query = query.Permute(ctx, 0, 2, 1, 3)
-	key = key.Permute(ctx, 0, 2, 1, 3)
-	value = value.Permute(ctx, 1, 2, 0, 3).Contiguous(ctx)
-	kq := key.MulmatFullPrec(ctx, query)
-	kq = kq.Scale(ctx, scale)
-	if mask != nil {
-		kq = kq.Add(ctx, mask)
-	}
-	kq = kq.Softmax(ctx)
-	kqv := value.Mulmat(ctx, kq)
-	attention := kqv.Permute(ctx, 0, 2, 1, 3).Contiguous(ctx)
-	attention = attention.Reshape(ctx, opts.hiddenSize, attention.Dim(2), batchSize)
-
-	return sa.Output.Forward(ctx, attention)
-}
-
-// VisionMLP implements the multi-layer perceptron
-type VisionMLP struct {
-	Gate *nn.Linear `gguf:"ffn_gate"`
-	Up   *nn.Linear `gguf:"ffn_up"`
-	Down *nn.Linear `gguf:"ffn_down"`
-}
-
-func (mlp *VisionMLP) Forward(ctx ml.Context, hiddenStates ml.Tensor, opts *VisionModelOptions) ml.Tensor {
-	// Using activation as specified in config (likely GELU or SiLU/Swish)
-	gateOutput := mlp.Gate.Forward(ctx, hiddenStates)
-	upOutput := mlp.Up.Forward(ctx, hiddenStates)
-	hiddenStates = gateOutput.SILU(ctx).Mul(ctx, upOutput)
-
-	return mlp.Down.Forward(ctx, hiddenStates)
-}
-
-type VisionEncoderLayer struct {
-	Norm1         *nn.RMSNorm `gguf:"ln1"`
-	SelfAttention *VisionSelfAttention
-	Norm2         *nn.RMSNorm `gguf:"ln2"`
-	MLP           *VisionMLP
-}
-
-func (e *VisionEncoderLayer) Forward(ctx ml.Context, hiddenStates, cos, sin, mask ml.Tensor, opts *VisionModelOptions) ml.Tensor {
-	residual := hiddenStates
-	hiddenStates = e.Norm1.Forward(ctx, hiddenStates, opts.eps)
-	hiddenStates = e.SelfAttention.Forward(ctx, hiddenStates, cos, sin, mask, opts)
-	hiddenStates = hiddenStates.Add(ctx, residual)
-
-	residual = hiddenStates
-	hiddenStates = e.Norm2.Forward(ctx, hiddenStates, opts.eps)
-	hiddenStates = e.MLP.Forward(ctx, hiddenStates, opts)
-	return hiddenStates.Add(ctx, residual)
-}
-
-// VisionModelOptions contains configuration options
-type VisionModelOptions struct {
-	hiddenSize        int
-	numHeads          int
-	headDim           int
-	patchSize         int
-	numChannels       int
-	eps               float32
-	ropeTheta         float32
-	spatialMergeSize  int
-	windowSize        int
-	fullAttnBlocks    []int32
-	temporalPatchSize int
-}
-
-type PatchEmbedding struct {
-	PatchConv0 *nn.Conv2D `gguf:"patch_embd_0"`
-	PatchConv1 *nn.Conv2D `gguf:"patch_embd_1"`
-}
-
-func (pe *PatchEmbedding) Forward(ctx ml.Context, pixelValues ml.Tensor, opts *VisionModelOptions) ml.Tensor {
-	numPatches := pixelValues.Shape()[1]
-
-	// Reshape the input tensor to match the expected dimensions
-	pixelValues = pixelValues.Reshape(ctx, opts.patchSize*opts.patchSize, opts.temporalPatchSize, opts.numChannels, numPatches)
-
-	// Permute the tensor to bring the temporal dimension to the front
-	pixelValues = pixelValues.Permute(ctx, 1, 0, 2, 3).Contiguous(ctx)
-
-	// Split the tensor into parts for the temporal convolutions
-	in0 := pixelValues.View(ctx, 0, 1, pixelValues.Stride(1), pixelValues.Dim(1), pixelValues.Stride(2), pixelValues.Dim(2), pixelValues.Stride(3), pixelValues.Dim(3)).Contiguous(ctx)
-	in0 = in0.Reshape(ctx, opts.patchSize, opts.patchSize, opts.numChannels, numPatches)
-	in1 := pixelValues.View(ctx, pixelValues.Stride(0), 1, pixelValues.Stride(1), pixelValues.Dim(1), pixelValues.Stride(2), pixelValues.Dim(2), pixelValues.Stride(3), pixelValues.Dim(3)).Contiguous(ctx)
-	in1 = in1.Reshape(ctx, opts.patchSize, opts.patchSize, opts.numChannels, numPatches)
-
-	s0, s1 := opts.patchSize, opts.patchSize // Use full stride
-	p0, p1 := 0, 0                           // padding
-	d0, d1 := 1, 1                           // dilation
-	out0 := pe.PatchConv0.Forward(ctx, in0, s0, s1, p0, p1, d0, d1)
-	out1 := pe.PatchConv1.Forward(ctx, in1, s0, s1, p0, p1, d0, d1)
-
-	// Add the outputs from the two temporal convolutions
-	out := out0.Add(ctx, out1)
-
-	// Reshape the output tensor to match the expected dimensions
-	return out.Reshape(ctx, opts.hiddenSize, numPatches)
-}
-
-// VisionPatchMerger implements patch merging for the Qwen vision model
-type VisionPatchMerger struct {
-	LNQ  *nn.RMSNorm `gguf:"ln_q"`
-	MLP0 *nn.Linear  `gguf:"mlp.0"`
-	MLP2 *nn.Linear  `gguf:"mlp.2"`
-}
-
-// Forward computes patch merging for the vision model
-func (pm *VisionPatchMerger) Forward(ctx ml.Context, visionOutputs ml.Tensor, opts *VisionModelOptions) ml.Tensor {
-	normalized := pm.LNQ.Forward(ctx, visionOutputs, opts.eps)
-
-	hiddenSize := visionOutputs.Dim(0) * (opts.spatialMergeSize * opts.spatialMergeSize)
-
-	// Reshape the normalized output to view the hidden size dimension
-	reshaped := normalized.Reshape(ctx, hiddenSize, normalized.Dim(1)/(opts.spatialMergeSize*opts.spatialMergeSize), batchSize)
-	hidden := pm.MLP0.Forward(ctx, reshaped)
-	activated := hidden.GELU(ctx)
-
-	output := pm.MLP2.Forward(ctx, activated)
-
-	return output
-}
-
-// VisionModel implements the Qwen vision model
-type VisionModel struct {
-	PatchEmbedding *PatchEmbedding
-	Layers         []VisionEncoderLayer `gguf:"blk"`
-	PatchMerger    *VisionPatchMerger   `gguf:"merger"`
-
-	*VisionModelOptions
-}
-
-// Forward computes the vision model for an input tensor
-func (m *VisionModel) Forward(ctx ml.Context, pixelValues ml.Tensor, grid *Grid) ml.Tensor {
-	// Extract patch embeddings
-	hiddenStates := m.PatchEmbedding.Forward(ctx, pixelValues, m.VisionModelOptions)
-
-	positionEmbedding := m.PositionalEmbedding(ctx, grid)
-
-	windowIndex, bounds := m.WindowIndex(ctx, grid)
-
-	spatialMergeUnit := m.spatialMergeSize * m.spatialMergeSize
-
-	hiddenStates = hiddenStates.Reshape(ctx, hiddenStates.Dim(0)*spatialMergeUnit, hiddenStates.Dim(1)/spatialMergeUnit)
-	hiddenStates = hiddenStates.Rows(ctx, windowIndex)
-	hiddenStates = hiddenStates.Reshape(ctx, hiddenStates.Dim(0)/spatialMergeUnit, hiddenStates.Dim(1)*spatialMergeUnit)
-
-	positionEmbedding = positionEmbedding.Reshape(ctx, positionEmbedding.Dim(0)*spatialMergeUnit, positionEmbedding.Dim(1)/spatialMergeUnit)
-	positionEmbedding = positionEmbedding.Rows(ctx, windowIndex)
-	positionEmbedding = positionEmbedding.Reshape(ctx, positionEmbedding.Dim(0)/spatialMergeUnit, positionEmbedding.Dim(1)*spatialMergeUnit)
-	positionEmbedding = positionEmbedding.Concat(ctx, positionEmbedding, 0)
-
-	cos, sin := positionEmbedding.Cos(ctx), positionEmbedding.Sin(ctx)
-	cos = cos.Reshape(ctx, cos.Dim(0), 1, cos.Dim(1))
-	sin = sin.Reshape(ctx, sin.Dim(0), 1, sin.Dim(1))
-
-	mask := blockDiagonalMask(ctx, hiddenStates.Dim(1), bounds, m.VisionModelOptions.numHeads)
-	// Apply encoder layers
-	for i, layer := range m.Layers {
-		if slices.Contains(m.fullAttnBlocks, int32(i)) {
-			hiddenStates = layer.Forward(ctx, hiddenStates, cos, sin, nil, m.VisionModelOptions)
-		} else {
-			hiddenStates = layer.Forward(
-				ctx,
-				hiddenStates,
-				cos,
-				sin,
-				mask,
-				m.VisionModelOptions,
-			)
-		}
-	}
-
-	hiddenStates = m.PatchMerger.Forward(ctx, hiddenStates, m.VisionModelOptions)
-	reverseWindowIndex := windowIndex.Argsort(ctx)
-	return hiddenStates.Rows(ctx, reverseWindowIndex)
-}
-
-// WindowIndex divides the grid into windows and returns:
-//  1. A tensor containing flattened indices of all grid points organized by windows
-//  2. A slice of boundaries that mark where each window's data begins and ends
-//     in the flattened representation, scaled by spatialMergeSize squared
-//
-// The boundaries slice always starts with 0 and contains cumulative ending
-// positions for each window, allowing downstream processing to identify
-// window boundaries in the tensor data.
-func (m *VisionModel) WindowIndex(ctx ml.Context, grid *Grid) (ml.Tensor, []int) {
-	vitMergerWindowSize := m.windowSize / m.spatialMergeSize / m.patchSize
-
-	llmGridH := grid.Height / m.spatialMergeSize
-	llmGridW := grid.Width / m.spatialMergeSize
-
-	// Calculate window parameters
-	numWindowsH := int(math.Ceil(float64(llmGridH) / float64(vitMergerWindowSize)))
-	numWindowsW := int(math.Ceil(float64(llmGridW) / float64(vitMergerWindowSize)))
-
-	// Initialize index_new slice
-	var index []int32
-
-	// Initialize bounds with the first element as 0
-	bounds := []int{0}
-	totalSeqLen := 0
-
-	// Process each window without padding
-	for wh := range numWindowsH {
-		for ww := range numWindowsW {
-			// Calculate window boundaries
-			hStart := wh * vitMergerWindowSize
-			wStart := ww * vitMergerWindowSize
-			hEnd := min(hStart+vitMergerWindowSize, llmGridH)
-			wEnd := min(wStart+vitMergerWindowSize, llmGridW)
-
-			// Calculate sequence length for this window
-			seqLen := (hEnd - hStart) * (wEnd - wStart)
-
-			// Collect indices for this window
-			for h := hStart; h < hEnd; h++ {
-				for w := wStart; w < wEnd; w++ {
-					index = append(index, int32(h*llmGridW+w))
-				}
-			}
-
-			totalSeqLen += seqLen
-			bounds = append(bounds, totalSeqLen*(m.spatialMergeSize*m.spatialMergeSize)+bounds[0])
-		}
-	}
-
-	t, err := ctx.Input().FromIntSlice(index, len(index))
-	if err != nil {
-		panic(err)
-	}
-
-	return t, bounds
-}
-
-// PositionalEmbedding generates rotary position embeddings for attention mechanisms
-func (m *VisionModel) PositionalEmbedding(ctx ml.Context, grid *Grid) ml.Tensor {
-	dim := m.headDim / 2
-	freq := dim / 2
-	theta := float64(m.ropeTheta)
-	merge := m.spatialMergeSize
-
-	// Create frequency patterns for position encoding
-	maxGridSize := max(grid.Height, grid.Width)
-	freqVals := make([]float32, freq*maxGridSize)
-	for i := range maxGridSize {
-		for j := range freq {
-			freqVals[i*freq+j] = float32(i) / float32(math.Pow(theta, float64(j*2)/float64(dim)))
-		}
-	}
-	freqs, err := ctx.Input().FromFloatSlice(freqVals, freq, maxGridSize)
-	if err != nil {
-		panic(fmt.Errorf("failed to create tensor from frequencies: %w", err))
-	}
-
-	// Create position coordinates (y,x pairs) for the grid
-	// In PyTorch: Equivalent to generating position ids with torch.arange()
-	coords := make([]int32, 0, grid.Height*grid.Width*2)
-	for y := range grid.Height {
-		for x := range grid.Width {
-			coords = append(coords, int32(y), int32(x))
-		}
-	}
-	pos, err := ctx.Input().FromIntSlice(coords, 2, grid.Width, grid.Height)
-	if err != nil {
-		panic(fmt.Errorf("failed to create tensor from positions: %w", err))
-	}
-
-	// Reshape and permute positions to match spatial merging pattern
-	pos = pos.Reshape(ctx, 2, grid.Width, merge, grid.Height/merge)
-	pos = pos.Permute(ctx, 0, 2, 1, 3).Contiguous(ctx)
-	pos = pos.Reshape(ctx, 2, merge, merge, grid.Width/merge*grid.Height/merge)
-	pos = pos.Permute(ctx, 0, 2, 1, 3).Contiguous(ctx)
-	pos = pos.Reshape(ctx, 2*merge*merge*grid.Width/merge*grid.Height/merge)
-
-	// Use position indices to look up corresponding frequency values
-	positionalEmbedding := freqs.Rows(ctx, pos)
-	positionalEmbedding = positionalEmbedding.Reshape(ctx, positionalEmbedding.Dim(0)*2, positionalEmbedding.Dim(1)/2)
-	return positionalEmbedding
-}
-
-// newVisionModel creates a new instance of the Qwen vision model
-func newVisionModel(c fs.Config) *VisionModel {
-	patchSize := int(c.Uint("vision.patch_size", 14))
-	hiddenSize := int(c.Uint("vision.embedding_length", 1280))
-	numHeads := int(c.Uint("vision.attention.head_count", 16))
-	numChannels := int(c.Uint("vision.num_channels", 3))
-	eps := c.Float("vision.attention.layer_norm_epsilon", 1e-6)
-	ropeTheta := c.Float("vision.rope.freq_base", 10000.0)
-	spatialMergeSize := int(c.Uint("vision.spatial_merge_size", 2))
-	windowSize := int(c.Uint("vision.window_size", 112))
-	fullAttnBlocks := c.Ints("qwen25vl.vision.fullatt_block_indexes", []int32{7, 15, 23, 31})
-	temporalPatchSize := int(c.Uint("vision.temporal_patch_size", 2))
-
-	model := &VisionModel{
-		Layers: make([]VisionEncoderLayer, c.Uint("vision.block_count", 32)),
-		VisionModelOptions: &VisionModelOptions{
-			hiddenSize:        hiddenSize,
-			numHeads:          numHeads,
-			headDim:           hiddenSize / numHeads,
-			patchSize:         patchSize,
-			numChannels:       numChannels,
-			eps:               eps,
-			ropeTheta:         ropeTheta,
-			spatialMergeSize:  spatialMergeSize,
-			windowSize:        windowSize,
-			temporalPatchSize: temporalPatchSize,
-			fullAttnBlocks:    fullAttnBlocks,
-		},
-	}
-
-	return model
-}

model/models/qwen25vl/process_image.go

@@ -1,184 +0,0 @@
-package qwen25vl
-
-import (
-	"fmt"
-	"image"
-	"math"
-
-	"github.com/ollama/ollama/fs"
-	"github.com/ollama/ollama/model/imageproc"
-)
-
-// ImageProcessor contains configuration for the Qwen 2.5 VL image processing
-type ImageProcessor struct {
-	numChannels       int
-	patchSize         int
-	temporalPatchSize int
-	mergeSize         int
-	minPixels         int
-	maxPixels         int
-	factor            int
-	rescaleFactor     float32
-	imageMean         []float32
-	imageStd          []float32
-}
-
-// newImageProcessor creates a new image processor with default values
-func newImageProcessor(c fs.Config) ImageProcessor {
-	patchSize := int(c.Uint("vision.patch_size", 14))
-	mergeSize := int(c.Uint("vision.spatial_merge_size", 2))
-
-	return ImageProcessor{
-		numChannels:       int(c.Uint("vision.num_channels", 3)), // not set
-		patchSize:         patchSize,
-		temporalPatchSize: 2,
-		mergeSize:         mergeSize,
-		minPixels:         56 * 56,
-		maxPixels:         int(c.Uint("vision.max_pixels", 28*28*1280)), // 1MP limit
-		factor:            patchSize * mergeSize,
-		rescaleFactor:     1.0 / 255.0,
-		imageMean:         imageproc.ClipDefaultMean[:],
-		imageStd:          imageproc.ClipDefaultSTD[:],
-	}
-}
-
-// SmartResize implements the smart resize algorithm
-func (p *ImageProcessor) SmartResize(height, width int) (int, int) {
-	factor := p.factor
-
-	if height < factor || width < factor {
-		panic(fmt.Sprintf("height:%d or width:%d must be larger than factor:%d", height, width, factor))
-	} else if aspectRatio := max(height, width) / min(height, width); aspectRatio > 200 {
-		panic(fmt.Sprintf("absolute aspect ratio must be smaller than 200, got %v", aspectRatio))
-	}
-
-	round := func(x float64) int { return int(math.RoundToEven(x)) }
-
-	hBar := round(float64(height)/float64(factor)) * factor
-	wBar := round(float64(width)/float64(factor)) * factor
-
-	if hBar*wBar > p.maxPixels {
-		beta := math.Sqrt(float64(height*width) / float64(p.maxPixels))
-
-		hBar = int(math.Floor(float64(height)/beta/float64(factor))) * factor
-		wBar = int(math.Floor(float64(width)/beta/float64(factor))) * factor
-	} else if hBar*wBar < p.minPixels {
-		beta := math.Sqrt(float64(p.minPixels) / float64(height*width))
-
-		hBar = int(math.Ceil(float64(height)*beta/float64(factor))) * factor
-		wBar = int(math.Ceil(float64(width)*beta/float64(factor))) * factor
-	}
-
-	return hBar, wBar
-}
-
-type Grid struct {
-	Height   int
-	Width    int
-	Temporal int
-}
-
-func (p *ImageProcessor) ProcessImage(img image.Image) ([]float32, *Grid, error) {
-	origWidth := img.Bounds().Dx()
-	origHeight := img.Bounds().Dy()
-
-	// Calculate smart resize dimensions
-	resizedHeight, resizedWidth := p.SmartResize(origHeight, origWidth)
-
-	// Resize image using existing functions
-	resizedImg := imageproc.Resize(img, image.Point{X: resizedWidth, Y: resizedHeight}, imageproc.ResizeBilinear)
-
-	normalizedPixels := imageproc.Normalize(
-		resizedImg,
-		[3]float32{p.imageMean[0], p.imageMean[1], p.imageMean[2]},
-		[3]float32{p.imageStd[0], p.imageStd[1], p.imageStd[2]},
-		true, // rescale
-		true, // channelFirst
-	)
-
-	// Calculate grid dimensions
-	grid := &Grid{
-		Height:   resizedHeight / p.patchSize,
-		Width:    resizedWidth / p.patchSize,
-		Temporal: 1, // For single images, temporal dimension is 1
-	}
-
-	patches, err := p.createPatches(normalizedPixels, resizedHeight, resizedWidth, grid)
-	if err != nil {
-		return nil, nil, fmt.Errorf("failed to create patches: %v", err)
-	}
-
-	// Return patches and grid dimensions
-	return patches, grid, nil
-}
-
-func (p *ImageProcessor) createPatches(pixels []float32, height, width int, grid *Grid) ([]float32, error) {
-	channels := p.numChannels
-	patchSize := p.patchSize
-	mergeSize := p.mergeSize
-	temporalPatchSize := p.temporalPatchSize
-
-	// Calculate output dimensions
-	numPatches := grid.Temporal * grid.Height * grid.Width
-	patchDim := channels * temporalPatchSize * patchSize * patchSize
-
-	result := make([]float32, numPatches*patchDim)
-	patchIndex := 0
-
-	// Single temporal frame handling (copies to all frames)
-	for range grid.Temporal {
-		for h := 0; h < grid.Height; h += mergeSize {
-			for w := 0; w < grid.Width; w += mergeSize {
-				// Handle the 2x2 merged patches
-				for mh := range mergeSize {
-					for mw := range mergeSize {
-						baseOffset := patchIndex * patchDim
-
-						// Extract patch data for first temporal frame
-						for c := range channels {
-							channelOffset := baseOffset + (c * temporalPatchSize * patchSize * patchSize)
-
-							for py := range patchSize {
-								for px := range patchSize {
-									// Calculate source pixel coordinates
-									y := (h+mh)*patchSize + py
-									x := (w+mw)*patchSize + px
-
-									// Source index in input tensor (CHW format)
-									srcIdx := c*height*width + y*width + x
-
-									// Destination index in first temporal frame
-									dstIdx := channelOffset + (py * patchSize) + px
-
-									if srcIdx < len(pixels) && dstIdx < len(result) {
-										result[dstIdx] = pixels[srcIdx]
-									}
-								}
-							}
-						}
-
-						// Copy first temporal frame to all other frames
-						if temporalPatchSize > 1 {
-							for c := range channels {
-								channelOffset := baseOffset + (c * temporalPatchSize * patchSize * patchSize)
-								firstFrameOffset := channelOffset
-								frameSize := patchSize * patchSize
-
-								// Copy first frame to all other frames
-								for tp := 1; tp < temporalPatchSize; tp++ {
-									currentFrameOffset := channelOffset + (tp * frameSize)
-									copy(result[currentFrameOffset:currentFrameOffset+frameSize],
-										result[firstFrameOffset:firstFrameOffset+frameSize])
-								}
-							}
-						}
-
-						patchIndex++
-					}
-				}
-			}
-		}
-	}
-
-	return result, nil
-}

runner/llamarunner/cache.go

@@ -104,8 +104,8 @@ func (c *InputCache) LoadCacheSlot(prompt []input, cachePrompt bool) (*InputCach
 	slog.Debug("loading cache slot", "id", slot.Id, "cache", len(slot.Inputs), "prompt", len(prompt),
 		"used", numPast, "remaining", len(prompt)-numPast)
 
-	slot.Inputs = prompt[:numPast]
 	prompt = prompt[numPast:]
+	slot.Inputs = slot.Inputs[:numPast]
 
 	return slot, prompt, nil
 }

runner/ollamarunner/cache.go

@@ -136,8 +136,8 @@ func (c *InputCache) LoadCacheSlot(prompt []input.Input) (*InputCacheSlot, []inp
 	slog.Debug("loading cache slot", "id", slot.Id, "cache", len(slot.Inputs), "prompt", len(prompt),
 		"used", numPast, "remaining", int32(len(prompt))-numPast)
 
-	slot.Inputs = prompt[:numPast]
 	prompt = prompt[numPast:]
+	slot.Inputs = slot.Inputs[:numPast]
 
 	return slot, prompt, nil
 }

runner/ollamarunner/cache_test.go

@@ -3,6 +3,7 @@ package ollamarunner
 import (
 	"errors"
 	"fmt"
+	"image"
 	"testing"
 	"time"
 
@@ -11,6 +12,10 @@ import (
 )
 
 func TestCountCommon(t *testing.T) {
+	imgA := image.NewRGBA(image.Rect(0, 0, 100, 100))
+	imgB := image.NewRGBA(image.Rect(0, 0, 50, 50))
+	imgC := image.NewRGBA(image.Rect(50, 50, 100, 100))
+
 	tests := []struct {
 		name     string
 		t1       []input.Input
@@ -31,20 +36,20 @@ func TestCountCommon(t *testing.T) {
 		},
 		{
 			name:     "Image Prefix",
-			t1:       []input.Input{{MultimodalHash: 1}},
-			t2:       []input.Input{{MultimodalHash: 1}, {MultimodalHash: 2}, {MultimodalHash: 3}},
+			t1:       []input.Input{{Multimodal: imgA, MultimodalHash: 1}},
+			t2:       []input.Input{{Multimodal: imgA, MultimodalHash: 1}, {Multimodal: imgB, MultimodalHash: 2}, {Multimodal: imgC, MultimodalHash: 3}},
 			expected: 1,
 		},
 		{
 			name:     "Mixed",
-			t1:       []input.Input{{Token: 1}, {MultimodalHash: 1}},
-			t2:       []input.Input{{Token: 1}, {MultimodalHash: 1}, {Token: 5}},
+			t1:       []input.Input{{Token: 1}, {Multimodal: imgA, MultimodalHash: 1}},
+			t2:       []input.Input{{Token: 1}, {Multimodal: imgA, MultimodalHash: 1}, {Token: 5}},
 			expected: 2,
 		},
 		{
 			name:     "Mixed, Same Length",
-			t1:       []input.Input{{Token: 1}, {MultimodalHash: 1}},
-			t2:       []input.Input{{Token: 1}, {MultimodalHash: 2}},
+			t1:       []input.Input{{Token: 1}, {Multimodal: imgA, MultimodalHash: 1}},
+			t2:       []input.Input{{Token: 1}, {Multimodal: imgB, MultimodalHash: 2}},
 			expected: 1,
 		},
 		{

runner/ollamarunner/multimodal.go

@@ -1,116 +0,0 @@
-package ollamarunner
-
-import (
-	"errors"
-
-	"github.com/ollama/ollama/ml"
-	"github.com/ollama/ollama/model/input"
-)
-
-// Tensors can't be used across multiple compute graphs. This is a problem
-// if a single embedding is split across batches using views since all of
-// the views will have the same source tensor. We also don't want to
-// recompute the entire embedding for each batch.
-//
-// To avoid this, we compute all of the tensors for the embedding on the
-// first use and then store the result in system memory. When we need
-// additional tensors, we recreate them from the stored data.
-
-// multimodalEntry represents the embeddings of a single object (such
-// as an image).
-type multimodalEntry struct {
-	// mm is the original set of tensors created by EncodeMultimodal
-	mm []input.Multimodal
-
-	// data is the computed result of mm. Nil if not yet computed
-	data [][]float32
-}
-
-// multimodalStore maps from an individual tensor (of which there
-// may be many in a single multimodal object) to its parent embedding
-type multimodalStore map[ml.Tensor]*multimodalEntry
-
-func newMultimodalStore() multimodalStore {
-	return make(multimodalStore)
-}
-
-// addMultimodal stores an embedding for later use in a compute graph
-func (m multimodalStore) addMultimodal(embedding []input.Multimodal) {
-	entry := &multimodalEntry{mm: embedding}
-
-	for _, e := range embedding {
-		if e.Tensor != nil {
-			m[e.Tensor] = entry
-		}
-	}
-}
-
-// getMultimodal takes a source set of tensors (which may contain a whole or
-// parts of one or more images) and returns the equivalent that can be used in
-// the current context
-func (m multimodalStore) getMultimodal(backend ml.Backend, ctx ml.Context, in []input.Multimodal, reserve bool) ([]input.Multimodal, error) {
-	out := make([]input.Multimodal, len(in))
-	for i := range out {
-		if in[i].Tensor != nil {
-			var err error
-			out[i].Tensor, err = m.getTensor(backend, ctx, in[i].Tensor, reserve)
-			if err != nil {
-				return nil, err
-			}
-		}
-
-		out[i].Data = in[i].Data
-	}
-
-	return out, nil
-}
-
-func (m multimodalStore) getTensor(backend ml.Backend, ctx ml.Context, in ml.Tensor, reserve bool) (ml.Tensor, error) {
-	entry := m[in]
-
-	if entry.data == nil {
-		computeCtx := backend.NewContext()
-		defer computeCtx.Close()
-
-		var tensors []ml.Tensor
-		for _, t := range entry.mm {
-			if t.Tensor != nil {
-				tensors = append(tensors, t.Tensor)
-			}
-		}
-
-		if len(tensors) == 0 {
-			return nil, nil
-		}
-
-		computeCtx.Forward(tensors...)
-		entry.data = make([][]float32, len(entry.mm))
-
-		if !reserve {
-			computeCtx.Compute(tensors...)
-
-			for i, t := range entry.mm {
-				if t.Tensor != nil {
-					entry.data[i] = t.Tensor.Floats()
-				}
-			}
-		} else {
-			err := computeCtx.Reserve()
-			if err != nil {
-				return nil, err
-			}
-		}
-	}
-
-	for i, t := range entry.mm {
-		if in == t.Tensor {
-			if !reserve {
-				return ctx.Input().FromFloatSlice(entry.data[i], t.Tensor.Shape()...)
-			} else {
-				return ctx.Input().Empty(t.Tensor.DType(), t.Tensor.Shape()...), nil
-			}
-		}
-	}
-
-	return nil, errors.New("multimodal tensor not found")
-}

runner/ollamarunner/runner.go

@@ -1,14 +1,12 @@
 package ollamarunner
 
 import (
-	"bytes"
 	"context"
 	"encoding/json"
 	"errors"
 	"flag"
 	"fmt"
 	"hash/maphash"
-	"image"
 	"log"
 	"log/slog"
 	"net"
@@ -22,7 +20,6 @@ import (
 	"time"
 	"unicode/utf8"
 
-	"golang.org/x/image/bmp"
 	"golang.org/x/sync/semaphore"
 
 	"github.com/ollama/ollama/api"
@@ -43,9 +40,6 @@ type Sequence struct {
 	// multimodal embeddings
 	ctxs []ml.Context
 
-	// mmStore holds multimodal embeddings to mange memory and enable splitting across batches
-	mmStore multimodalStore
-
 	// batch index
 	iBatch int
 
@@ -107,7 +101,7 @@ func (s *Server) NewSequence(prompt string, images []llm.ImageData, params NewSe
 
 	startTime := time.Now()
 
-	inputs, ctxs, mmStore, err := s.inputs(prompt, images)
+	inputs, ctxs, err := s.inputs(prompt, images)
 	if err != nil {
 		return nil, fmt.Errorf("failed to process inputs: %w", err)
 	} else if len(inputs) == 0 {
@@ -162,7 +156,6 @@ func (s *Server) NewSequence(prompt string, images []llm.ImageData, params NewSe
 
 	return &Sequence{
 		ctxs:                ctxs,
-		mmStore:             mmStore,
 		inputs:              inputs,
 		numPromptInputs:     len(inputs),
 		startProcessingTime: startTime,
@@ -181,10 +174,9 @@ func (s *Server) NewSequence(prompt string, images []llm.ImageData, params NewSe
 // inputs processes the prompt and images into a list of inputs
 // by splitting the prompt on [img-<n>] tags, tokenizing text and
 // decoding images
-func (s *Server) inputs(prompt string, images []llm.ImageData) ([]input.Input, []ml.Context, multimodalStore, error) {
+func (s *Server) inputs(prompt string, images []llm.ImageData) ([]input.Input, []ml.Context, error) {
 	var inputs []input.Input
 	var ctxs []ml.Context
-	var mmStore multimodalStore
 
 	var parts []string
 	var matches [][]string
@@ -195,7 +187,6 @@ func (s *Server) inputs(prompt string, images []llm.ImageData) ([]input.Input, [
 		re := regexp.MustCompile(`\[img-(\d+)\]`)
 		parts = re.Split(prompt, -1)
 		matches = re.FindAllStringSubmatch(prompt, -1)
-		mmStore = newMultimodalStore()
 	} else {
 		parts = []string{prompt}
 	}
@@ -205,7 +196,7 @@ func (s *Server) inputs(prompt string, images []llm.ImageData) ([]input.Input, [
 		// text - tokenize
 		tokens, err := s.model.(model.TextProcessor).Encode(part, i == 0)
 		if err != nil {
-			return nil, nil, nil, err
+			return nil, nil, err
 		}
 
 		for _, t := range tokens {
@@ -225,7 +216,7 @@ func (s *Server) inputs(prompt string, images []llm.ImageData) ([]input.Input, [
 			}
 
 			if imageIndex < 0 {
-				return nil, nil, nil, fmt.Errorf("invalid image index: %d", n)
+				return nil, nil, fmt.Errorf("invalid image index: %d", n)
 			}
 
 			ctx := s.model.Backend().NewContext()
@@ -233,15 +224,13 @@ func (s *Server) inputs(prompt string, images []llm.ImageData) ([]input.Input, [
 			ctxs = append(ctxs, ctx)
 			imageEmbeddings, err := multimodalProcessor.EncodeMultimodal(ctx, images[imageIndex].Data)
 			if err != nil {
-				return nil, nil, nil, err
+				return nil, nil, err
 			}
 
 			s.multimodalHash.Reset()
 			_, _ = s.multimodalHash.Write(images[imageIndex].Data)
 			imageHash := s.multimodalHash.Sum64()
 
-			mmStore.addMultimodal(imageEmbeddings)
-
 			inputs = append(inputs, input.Input{Multimodal: imageEmbeddings, MultimodalHash: imageHash})
 			postTokenize = true
 		}
@@ -251,11 +240,11 @@ func (s *Server) inputs(prompt string, images []llm.ImageData) ([]input.Input, [
 		var err error
 		inputs, err = multimodalProcessor.PostTokenize(inputs)
 		if err != nil {
-			return nil, nil, nil, err
+			return nil, nil, err
 		}
 	}
 
-	return inputs, ctxs, mmStore, nil
+	return inputs, ctxs, nil
 }
 
 type Server struct {
@@ -374,9 +363,6 @@ func (s *Server) processBatch() error {
 	}
 	defer s.mu.Unlock()
 
-	ctx := s.model.Backend().NewContext()
-	defer ctx.Close()
-
 	var batchInputs []int32
 	var batch input.Batch
 
@@ -447,11 +433,7 @@ func (s *Server) processBatch() error {
 
 			batchInputs = append(batchInputs, inp.Token)
 			if inp.Multimodal != nil {
-				mm, err := seq.mmStore.getMultimodal(s.model.Backend(), ctx, inp.Multimodal, false)
-				if err != nil {
-					return err
-				}
-				batch.Multimodal = append(batch.Multimodal, input.MultimodalIndex{Index: len(batchInputs) - 1, Multimodal: mm})
+				batch.Multimodal = append(batch.Multimodal, input.MultimodalIndex{Index: len(batchInputs) - 1, Multimodal: inp.Multimodal})
 			}
 
 			batch.Positions = append(batch.Positions, int32(len(seq.cache.Inputs)+len(seq.pendingInputs)))
@@ -477,6 +459,9 @@ func (s *Server) processBatch() error {
 		return nil
 	}
 
+	ctx := s.model.Backend().NewContext()
+	defer ctx.Close()
+
 	modelOutput, err := model.Forward(ctx, s.model, batchInputs, batch)
 	if err != nil {
 		return fmt.Errorf("failed to decode batch: %w", err)
@@ -735,71 +720,12 @@ func (s *Server) reserveWorstCaseGraph() error {
 	ctx := s.model.Backend().NewContext()
 	defer ctx.Close()
 
-	var err error
-	inputs := make([]input.Input, s.batchSize)
-	mmStore := newMultimodalStore()
-
-	// Multimodal strategy:
-	// - Encode a 2048x2048 image. This assumes that a single image of this
-	//   size is sufficient to trigger the worst case. This is currently true
-	//   because for existing models, only a single image fits in a batch.
-	// - Add the embedding to a full batch of tokens - this is necessary because
-	//   the model may be looking for non-image data, such as <image> tags.
-	// - Run PostTokenize to execute any transformations between generated
-	//   embeddings and what the forward pass expects.
-	// - The result may now be larger than a batch (images may not fit in a
-	//   single batch), so trim based on what will fit and must be grouped together.
-	// - Fill out the rest of the space with text tokens.
-	if multimodalProcessor, ok := s.model.(model.MultimodalProcessor); ok {
-		mmCtx := s.model.Backend().NewContext()
-		defer mmCtx.Close()
-
-		img := image.NewGray(image.Rect(0, 0, 2048, 2048))
-		var buf bytes.Buffer
-		bmp.Encode(&buf, img)
-
-		if inputs[0].Multimodal, err = multimodalProcessor.EncodeMultimodal(mmCtx, buf.Bytes()); err == nil {
-			mmStore.addMultimodal(inputs[0].Multimodal)
-
-			inputs, err = multimodalProcessor.PostTokenize(inputs)
-			if err != nil {
-				return err
-			}
-
-			for i, inp := range inputs {
-				minBatch := 1 + inp.SameBatch
-				if minBatch > s.batchSize {
-					inputs = inputs[i:min(i+minBatch, len(inputs))]
-					break
-				} else if i+minBatch > s.batchSize {
-					inputs = inputs[:i]
-					break
-				}
-			}
-
-			if len(inputs) < s.batchSize {
-				newInputs := make([]input.Input, s.batchSize)
-				copy(newInputs, inputs)
-				inputs = newInputs
-			}
-		}
-	}
-
 	var batch input.Batch
 
-	batchInputs := make([]int32, len(inputs))
+	inputs := make([]int32, s.batchSize)
 	batch.Positions = make([]int32, len(inputs))
 	batch.Sequences = make([]int, len(inputs))
-	for i, inp := range inputs {
-		batchInputs[i] = inp.Token
-		if inp.Multimodal != nil {
-			mm, err := mmStore.getMultimodal(s.model.Backend(), ctx, inp.Multimodal, true)
-			if err != nil {
-				return err
-			}
-			batch.Multimodal = append(batch.Multimodal, input.MultimodalIndex{Index: i, Multimodal: mm})
-		}
-
+	for i := range inputs {
 		batch.Positions[i] = int32(i)
 	}
 
@@ -808,7 +734,8 @@ func (s *Server) reserveWorstCaseGraph() error {
 		batch.Outputs[i] = int32(i)
 	}
 
-	batch.Inputs, err = ctx.Input().FromIntSlice(batchInputs, len(batchInputs))
+	var err error
+	batch.Inputs, err = ctx.Input().FromIntSlice(inputs, len(inputs))
 	if err != nil {
 		return err
 	}

server/create.go

@@ -430,7 +430,7 @@ func quantizeLayer(layer *layerGGML, quantizeType string, fn func(resp api.Progr
 	fnWrap := func(n uint64) {
 		done := doneBytes.Add(n)
 		progress := float32(done) / float32(totalBytes)
-		fn(api.ProgressResponse{Status: fmt.Sprintf("quantizing %s model to %s", ft, quantizeType), Digest: "0000000000000000000", Total: layer.Size, Completed: int64(progress * float32(layer.Size))})
+		fn(api.ProgressResponse{Status: fmt.Sprintf("quantizing %s model to %s", ft, quantizeType), Digest: "0", Total: layer.Size, Completed: int64(progress * float32(layer.Size))})
 	}
 	ftype, err := ggml.ParseFileType(quantizeType)
 	if err != nil {
@@ -467,7 +467,7 @@ func quantizeLayer(layer *layerGGML, quantizeType string, fn func(resp api.Progr
 		return nil, err
 	}
 
-	f, _, err := ggml.Decode(temp, 0)
+	f, _, err := ggml.Decode(temp, 1024)
 	if err != nil {
 		slog.Error(fmt.Sprintf("error decoding ggml: %s\n", err))
 		return nil, err

server/images.go

@@ -75,7 +75,7 @@ func (m *Model) Capabilities() []model.Capability {
 	if err == nil {
 		defer r.Close()
 
-		f, _, err := ggml.Decode(r, 0)
+		f, _, err := ggml.Decode(r, 1024)
 		if err == nil {
 			if _, ok := f.KV()[fmt.Sprintf("%s.pooling_type", f.KV().Architecture())]; ok {
 				capabilities = append(capabilities, model.CapabilityEmbedding)

server/model.go

@@ -10,9 +10,6 @@ import (
 	"log/slog"
 	"net/http"
 	"os"
-	"slices"
-	"strings"
-	"text/template/parse"
 
 	"github.com/ollama/ollama/api"
 	"github.com/ollama/ollama/fs/ggml"
@@ -128,124 +125,3 @@ func detectContentType(r io.Reader) (string, error) {
 
 	return "unknown", nil
 }
-
-func parseObjects(s string) []map[string]any {
-	var objs []map[string]any
-	for offset := 0; offset < len(s); {
-		var obj map[string]any
-		decoder := json.NewDecoder(strings.NewReader(s[offset:]))
-		if err := decoder.Decode(&obj); errors.Is(err, io.EOF) || errors.Is(err, io.ErrUnexpectedEOF) {
-			break
-		} else if syntax := &(json.SyntaxError{}); errors.As(err, &syntax) {
-			// skip over any syntax errors
-			offset += int(syntax.Offset)
-		} else if unmarshalType := &(json.UnmarshalTypeError{}); errors.As(err, &unmarshalType) {
-			// skip over any unmarshalable types
-			offset += int(unmarshalType.Offset)
-		} else if err != nil {
-			return nil
-		} else {
-			offset += int(decoder.InputOffset())
-			objs = append(objs, obj)
-		}
-	}
-
-	return objs
-}
-
-// parseToolCalls attempts to parse a JSON string into a slice of ToolCalls.
-// mxyng: this only really works if the input contains tool calls in some JSON format
-func (m *Model) parseToolCalls(s string) ([]api.ToolCall, bool) {
-	// create a subtree from the node that ranges over .ToolCalls
-	tmpl := m.Template.Subtree(func(n parse.Node) bool {
-		if t, ok := n.(*parse.RangeNode); ok {
-			return slices.Contains(template.Identifiers(t.Pipe), "ToolCalls")
-		}
-
-		return false
-	})
-
-	if tmpl == nil {
-		return nil, false
-	}
-
-	var b bytes.Buffer
-	if err := tmpl.Execute(&b, map[string][]api.ToolCall{
-		"ToolCalls": {
-			{
-				Function: api.ToolCallFunction{
-					Name: "@@name@@",
-					Arguments: api.ToolCallFunctionArguments{
-						"@@argument@@": 1,
-					},
-				},
-			},
-		},
-	}); err != nil {
-		return nil, false
-	}
-
-	templateObjects := parseObjects(b.String())
-	if len(templateObjects) == 0 {
-		return nil, false
-	}
-
-	// find the keys that correspond to the name and arguments fields
-	var name, arguments string
-	for k, v := range templateObjects[0] {
-		switch v.(type) {
-		case string:
-			name = k
-		case map[string]any:
-			arguments = k
-		}
-	}
-
-	if name == "" || arguments == "" {
-		return nil, false
-	}
-
-	responseObjects := parseObjects(s)
-	if len(responseObjects) == 0 {
-		return nil, false
-	}
-
-	// collect all nested objects
-	var collect func(any) []map[string]any
-	collect = func(obj any) (all []map[string]any) {
-		switch o := obj.(type) {
-		case map[string]any:
-			all = append(all, o)
-			for _, v := range o {
-				all = append(all, collect(v)...)
-			}
-		case []any:
-			for _, v := range o {
-				all = append(all, collect(v)...)
-			}
-		}
-
-		return all
-	}
-
-	var objs []map[string]any
-	for _, p := range responseObjects {
-		objs = append(objs, collect(p)...)
-	}
-
-	var toolCalls []api.ToolCall
-	for _, kv := range objs {
-		n, nok := kv[name].(string)
-		a, aok := kv[arguments].(map[string]any)
-		if nok && aok {
-			toolCalls = append(toolCalls, api.ToolCall{
-				Function: api.ToolCallFunction{
-					Name:      n,
-					Arguments: a,
-				},
-			})
-		}
-	}
-
-	return toolCalls, len(toolCalls) > 0
-}

server/model_test.go

@@ -1,179 +0,0 @@
-package server
-
-import (
-	"bytes"
-	"encoding/json"
-	"fmt"
-	"os"
-	"path/filepath"
-	"testing"
-
-	"github.com/google/go-cmp/cmp"
-
-	"github.com/ollama/ollama/api"
-	"github.com/ollama/ollama/template"
-)
-
-func readFile(t *testing.T, base, name string) *bytes.Buffer {
-	t.Helper()
-
-	bts, err := os.ReadFile(filepath.Join(base, name))
-	if err != nil {
-		t.Fatal(err)
-	}
-
-	return bytes.NewBuffer(bts)
-}
-
-func TestExecuteWithTools(t *testing.T) {
-	p := filepath.Join("testdata", "tools")
-	cases := []struct {
-		model  string
-		output string
-		ok     bool
-	}{
-		{"mistral", `[TOOL_CALLS]  [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`, true},
-		{"mistral", `[TOOL_CALLS]  [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]
-
-The temperature in San Francisco, CA is 70°F and in Toronto, Canada is 20°C.`, true},
-		{"mistral", `[TOOL_CALLS]  [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"To }]`, false},
-		{"mistral", `I'm not aware of that information. However, I can suggest searching for the weather using the "get_current_weather" function:
-
-		[{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`, true},
-		{"mistral", " The weather in San Francisco, CA is 70°F and in Toronto, Canada is 20°C.", false},
-		{"command-r-plus", "Action: ```json" + `
-[
-    {
-        "tool_name": "get_current_weather",
-        "parameters": {
-            "format": "fahrenheit",
-            "location": "San Francisco, CA"
-        }
-    },
-    {
-        "tool_name": "get_current_weather",
-        "parameters": {
-            "format": "celsius",
-            "location": "Toronto, Canada"
-        }
-    }
-]
-` + "```", true},
-		{"command-r-plus", " The weather in San Francisco, CA is 70°F and in Toronto, Canada is 20°C.", false},
-		{"firefunction", ` functools[{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`, true},
-		{"firefunction", " The weather in San Francisco, CA is 70°F and in Toronto, Canada is 20°C.", false},
-		{"llama3-groq-tool-use", `<tool_call>
-{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}
-{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}
-</tool_call>`, true},
-		{"xlam", `{"tool_calls": [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]}`, true},
-		{"nemotron", `<toolcall>{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]} </toolcall>`, true},
-	}
-
-	var tools []api.Tool
-	if err := json.Unmarshal(readFile(t, p, "tools.json").Bytes(), &tools); err != nil {
-		t.Fatal(err)
-	}
-
-	var messages []api.Message
-	if err := json.Unmarshal(readFile(t, p, "messages.json").Bytes(), &messages); err != nil {
-		t.Fatal(err)
-	}
-
-	calls := []api.ToolCall{
-		{
-			Function: api.ToolCallFunction{
-				Name: "get_current_weather",
-				Arguments: api.ToolCallFunctionArguments{
-					"format":   "fahrenheit",
-					"location": "San Francisco, CA",
-				},
-			},
-		},
-		{
-			Function: api.ToolCallFunction{
-				Name: "get_current_weather",
-				Arguments: api.ToolCallFunctionArguments{
-					"format":   "celsius",
-					"location": "Toronto, Canada",
-				},
-			},
-		},
-	}
-
-	for _, tt := range cases {
-		t.Run(tt.model, func(t *testing.T) {
-			tmpl, err := template.Parse(readFile(t, p, fmt.Sprintf("%s.gotmpl", tt.model)).String())
-			if err != nil {
-				t.Fatal(err)
-			}
-
-			t.Run("template", func(t *testing.T) {
-				var actual bytes.Buffer
-				if err := tmpl.Execute(&actual, template.Values{Tools: tools, Messages: messages}); err != nil {
-					t.Fatal(err)
-				}
-
-				if diff := cmp.Diff(actual.String(), readFile(t, p, fmt.Sprintf("%s.out", tt.model)).String()); diff != "" {
-					t.Errorf("mismatch (-got +want):\n%s", diff)
-				}
-			})
-
-			t.Run("parse", func(t *testing.T) {
-				m := &Model{Template: tmpl}
-				actual, ok := m.parseToolCalls(tt.output)
-				if ok != tt.ok {
-					t.Fatalf("expected %t, got %t", tt.ok, ok)
-				}
-
-				if tt.ok {
-					if diff := cmp.Diff(actual, calls); diff != "" {
-						t.Errorf("mismatch (-got +want):\n%s", diff)
-					}
-				}
-			})
-		})
-	}
-}
-
-func TestParseObjects(t *testing.T) {
-	tests := []struct {
-		input string
-		want  []map[string]any
-	}{
-		{
-			input: `[{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
-			want: []map[string]any{
-				{"name": "get_current_weather", "arguments": map[string]any{"format": "fahrenheit", "location": "San Francisco, CA"}},
-				{"name": "get_current_weather", "arguments": map[string]any{"format": "celsius", "location": "Toronto, Canada"}},
-			},
-		},
-		{
-			input: `<toolcall>{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} </toolcall>`,
-			want: []map[string]any{
-				{"name": "get_current_weather", "arguments": map[string]any{"format": "fahrenheit", "location": "San Francisco, CA"}},
-			},
-		},
-		{
-			input: `<toolcall>{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} </toolcall> <toolcall>{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, ON"}} </toolcall>`,
-			want: []map[string]any{
-				{"name": "get_current_weather", "arguments": map[string]any{"format": "fahrenheit", "location": "San Francisco, CA"}},
-				{"name": "get_current_weather", "arguments": map[string]any{"format": "celsius", "location": "Toronto, ON"}},
-			},
-		},
-		{
-			input: `{"name": "get_current_weather", "arguments": `,
-			want:  nil,
-		},
-	}
-
-	for _, tc := range tests {
-		t.Run(tc.input, func(t *testing.T) {
-			got := parseObjects(tc.input)
-
-			if diff := cmp.Diff(got, tc.want); diff != "" {
-				t.Errorf("mismatch (-got +want):\n%s", diff)
-			}
-		})
-	}
-}

server/routes.go

@@ -38,6 +38,7 @@ import (
 	"github.com/ollama/ollama/server/internal/client/ollama"
 	"github.com/ollama/ollama/server/internal/registry"
 	"github.com/ollama/ollama/template"
+	"github.com/ollama/ollama/tools"
 	"github.com/ollama/ollama/types/errtypes"
 	"github.com/ollama/ollama/types/model"
 	"github.com/ollama/ollama/version"
@@ -1482,11 +1483,20 @@ func (s *Server) ChatHandler(c *gin.Context) {
 		return
 	}
 
+	var toolParser tools.Parser
+	if len(req.Tools) > 0 {
+		toolParser, err = tools.NewParser(m.Template.Template)
+		if err != nil {
+			slog.Error("failed to create tool parser", "error", err)
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+	}
+
 	ch := make(chan any)
 	go func() {
 		defer close(ch)
-		var sb strings.Builder
-		var toolCallIndex int = 0
+
 		if err := r.Completion(c.Request.Context(), llm.CompletionRequest{
 			Prompt:  prompt,
 			Images:  images,
@@ -1512,37 +1522,26 @@ func (s *Server) ChatHandler(c *gin.Context) {
 				res.LoadDuration = checkpointLoaded.Sub(checkpointStart)
 			}
 
-			// TODO: tool call checking and filtering should be moved outside of this callback once streaming
-			// however this was a simple change for now without reworking streaming logic of this (and other)
-			// handlers
-			if req.Stream != nil && !*req.Stream || len(req.Tools) == 0 {
-				ch <- res
-				return
-			}
-
-			// Streaming tool calls:
-			// If tools are recognized, use a flag to track the sending of a tool downstream
-			// This ensures that content is cleared from the message on the last chunk sent
-			sb.WriteString(r.Content)
-			if toolCalls, ok := m.parseToolCalls(sb.String()); ok {
-				res.Message.ToolCalls = toolCalls
-				for i := range toolCalls {
-					toolCalls[i].Function.Index = toolCallIndex
-					toolCallIndex++
+			if len(req.Tools) > 0 && !toolParser.Done {
+				if r.Content == "" {
+					return
 				}
-				res.Message.Content = ""
-				sb.Reset()
-				ch <- res
-				return
-			}
-
-			if r.Done {
-				// Send any remaining content if no tool calls were detected
-				if toolCallIndex == 0 {
-					res.Message.Content = sb.String()
+				toolCalls, content, err := toolParser.Add(r.Content)
+				if err == nil {
+					if len(content) > 0 {
+						res.Message.Content = content
+						slog.Debug("tools: setting content to", "content", content)
+					} else if len(toolCalls) > 0 {
+						res.Message.ToolCalls = toolCalls
+						res.Message.Content = ""
+					}
+				} else if errors.Is(err, tools.ErrAccumulateMore) {
+					return
+				} else {
+					slog.Debug("tools: error", "error", err)
 				}
-				ch <- res
 			}
+			ch <- res
 		}); err != nil {
 			ch <- gin.H{"error": err.Error()}
 		}
@@ -1551,11 +1550,15 @@ func (s *Server) ChatHandler(c *gin.Context) {
 	if req.Stream != nil && !*req.Stream {
 		var resp api.ChatResponse
 		var sb strings.Builder
+		var toolCalls []api.ToolCall
 		for rr := range ch {
 			switch t := rr.(type) {
 			case api.ChatResponse:
 				sb.WriteString(t.Message.Content)
 				resp = t
+				if len(req.Tools) > 0 {
+					toolCalls = append(toolCalls, t.Message.ToolCalls...)
+				}
 			case gin.H:
 				msg, ok := t["error"].(string)
 				if !ok {
@@ -1571,12 +1574,8 @@ func (s *Server) ChatHandler(c *gin.Context) {
 		}
 
 		resp.Message.Content = sb.String()
-
-		if len(req.Tools) > 0 {
-			if toolCalls, ok := m.parseToolCalls(sb.String()); ok {
-				resp.Message.ToolCalls = toolCalls
-				resp.Message.Content = ""
-			}
+		if len(toolCalls) > 0 {
+			resp.Message.ToolCalls = toolCalls
 		}
 
 		c.JSON(http.StatusOK, resp)

tools/testdata/llama3.2.gotmpl

@@ -0,0 +1,44 @@
+<|start_header_id|>system<|end_header_id|>
+
+Cutting Knowledge Date: December 2023
+
+{{ if .System }}{{ .System }}
+{{- end }}
+{{- if .Tools }}When you receive a tool call response, use the output to format an answer to the orginal user question.
+
+You are a helpful assistant with tool calling capabilities.
+{{- end }}<|eot_id|>
+{{- range $i, $_ := .Messages }}
+{{- $last := eq (len (slice $.Messages $i)) 1 }}
+{{- if eq .Role "user" }}<|start_header_id|>user<|end_header_id|>
+{{- if and $.Tools $last }}
+
+Given the following functions, please respond with a JSON for a function call with its proper arguments that best answers the given prompt.
+
+Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}. Do not use variables.
+
+{{ range $.Tools }}
+{{- . }}
+{{ end }}
+{{ .Content }}<|eot_id|>
+{{- else }}
+
+{{ .Content }}<|eot_id|>
+{{- end }}{{ if $last }}<|start_header_id|>assistant<|end_header_id|>
+
+{{ end }}
+{{- else if eq .Role "assistant" }}<|start_header_id|>assistant<|end_header_id|>
+{{- if .ToolCalls }}
+{{ range .ToolCalls }}
+{"name": "{{ .Function.Name }}", "parameters": {{ .Function.Arguments }}}{{ end }}
+{{- else }}
+
+{{ .Content }}
+{{- end }}{{ if not $last }}<|eot_id|>{{ end }}
+{{- else if eq .Role "tool" }}<|start_header_id|>ipython<|end_header_id|>
+
+{{ .Content }}<|eot_id|>{{ if $last }}<|start_header_id|>assistant<|end_header_id|>
+
+{{ end }}
+{{- end }}
+{{- end }}

tools/testdata/llama3.2.out

@@ -0,0 +1,24 @@
+<|start_header_id|>system<|end_header_id|>
+
+Cutting Knowledge Date: December 2023
+
+You are a knowledgeable assistant. You can answer questions and perform tasks.When you receive a tool call response, use the output to format an answer to the orginal user question.
+
+You are a helpful assistant with tool calling capabilities.<|eot_id|><|start_header_id|>user<|end_header_id|>
+
+What's the weather like today in Paris?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
+
+{"name": "get_current_weather", "parameters": {"format":"celsius","location":"Paris, France"}}<|eot_id|><|start_header_id|>ipython<|end_header_id|>
+
+22<|eot_id|><|start_header_id|>assistant<|end_header_id|>
+
+The current temperature in Paris, France is 22 degrees Celsius.<|eot_id|><|start_header_id|>user<|end_header_id|>
+
+Given the following functions, please respond with a JSON for a function call with its proper arguments that best answers the given prompt.
+
+Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}. Do not use variables.
+
+{"type":"function","function":{"name":"get_current_weather","description":"Get the current weather","parameters":{"type":"object","required":["location","format"],"properties":{"format":{"type":"string","description":"The temperature unit to use. Infer this from the user's location.","enum":["celsius","fahrenheit"]},"location":{"type":"string","description":"The city and state, e.g. San Francisco, CA"}}}}}
+
+What's the weather like today in San Francisco and Toronto?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
+

tools/testdata/qwen2.5-coder.gotmpl

@@ -0,0 +1,51 @@
+{{- if .Suffix }}<|fim_prefix|>{{ .Prompt }}<|fim_suffix|>{{ .Suffix }}<|fim_middle|>
+{{- else if .Messages }}
+{{- if or .System .Tools }}<|im_start|>system
+{{- if .System }}
+{{ .System }}
+{{- end }}
+{{- if .Tools }}
+
+# Tools
+
+You may call one or more functions to assist with the user query.
+
+You are provided with function signatures within <tools></tools> XML tags:
+<tools>
+{{- range .Tools }}
+{"type": "function", "function": {{ .Function }}}
+{{- end }}
+</tools>
+
+For each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:
+<tool_call>
+{"name": <function-name>, "arguments": <args-json-object>}
+</tool_call>
+{{- end }}<|im_end|>
+{{ end }}
+{{- range $i, $_ := .Messages }}
+{{- $last := eq (len (slice $.Messages $i)) 1 -}}
+{{- if eq .Role "user" }}<|im_start|>user
+{{ .Content }}<|im_end|>
+{{ else if eq .Role "assistant" }}<|im_start|>assistant
+{{ if .Content }}{{ .Content }}
+{{- else if .ToolCalls }}<tool_call>
+{{ range .ToolCalls }}{"name": "{{ .Function.Name }}", "arguments": {{ .Function.Arguments }}}
+{{ end }}</tool_call>
+{{- end }}{{ if not $last }}<|im_end|>
+{{ end }}
+{{- else if eq .Role "tool" }}<|im_start|>user
+<tool_response>
+{{ .Content }}
+</tool_response><|im_end|>
+{{ end }}
+{{- if and (ne .Role "assistant") $last }}<|im_start|>assistant
+{{ end }}
+{{- end }}
+{{- else }}
+{{- if .System }}<|im_start|>system
+{{ .System }}<|im_end|>
+{{ end }}{{ if .Prompt }}<|im_start|>user
+{{ .Prompt }}<|im_end|>
+{{ end }}<|im_start|>assistant
+{{ end }}{{ .Response }}{{ if .Response }}<|im_end|>{{ end }}

tools/testdata/qwen2.5-coder.out

@@ -0,0 +1,31 @@
+<|im_start|>system
+You are a knowledgeable assistant. You can answer questions and perform tasks.
+
+# Tools
+
+You may call one or more functions to assist with the user query.
+
+You are provided with function signatures within <tools></tools> XML tags:
+<tools>
+{"type": "function", "function": {"name":"get_current_weather","description":"Get the current weather","parameters":{"type":"object","required":["location","format"],"properties":{"format":{"type":"string","description":"The temperature unit to use. Infer this from the user's location.","enum":["celsius","fahrenheit"]},"location":{"type":"string","description":"The city and state, e.g. San Francisco, CA"}}}}}
+</tools>
+
+For each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:
+<tool_call>
+{"name": <function-name>, "arguments": <args-json-object>}
+</tool_call><|im_end|>
+<|im_start|>user
+What's the weather like today in Paris?<|im_end|>
+<|im_start|>assistant
+<tool_call>
+{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Paris, France"}}
+</tool_call><|im_end|>
+<|im_start|>user
+<tool_response>
+22
+</tool_response><|im_end|>
+<|im_start|>assistant
+The current temperature in Paris, France is 22 degrees Celsius.<|im_end|>
+<|im_start|>user
+What's the weather like today in San Francisco and Toronto?<|im_end|>
+<|im_start|>assistant

tools/testdata/qwen3.gotmpl

@@ -0,0 +1,50 @@
+{{- if .Messages }}
+{{- if or .System .Tools }}<|im_start|>system
+{{- if .System }}
+{{ .System }}
+{{- end }}
+{{- if .Tools }}
+
+# Tools
+
+You may call one or more functions to assist with the user query.
+
+You are provided with function signatures within <tools></tools> XML tags:
+<tools>
+{{- range .Tools }}
+{"type": "function", "function": {{ .Function }}}
+{{- end }}
+</tools>
+
+For each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:
+<tool_call>
+{"name": <function-name>, "arguments": <args-json-object>}
+</tool_call>
+{{- end }}<|im_end|>
+{{ end }}
+{{- range $i, $_ := .Messages }}
+{{- $last := eq (len (slice $.Messages $i)) 1 -}}
+{{- if eq .Role "user" }}<|im_start|>user
+{{ .Content }}<|im_end|>
+{{ else if eq .Role "assistant" }}<|im_start|>assistant
+{{ if .Content }}{{ .Content }}
+{{- else if .ToolCalls }}<tool_call>
+{{ range .ToolCalls }}{"name": "{{ .Function.Name }}", "arguments": {{ .Function.Arguments }}}
+{{ end }}</tool_call>
+{{- end }}{{ if not $last }}<|im_end|>
+{{ end }}
+{{- else if eq .Role "tool" }}<|im_start|>user
+<tool_response>
+{{ .Content }}
+</tool_response><|im_end|>
+{{ end }}
+{{- if and (ne .Role "assistant") $last }}<|im_start|>assistant
+{{ end }}
+{{- end }}
+{{- else }}
+{{- if .System }}<|im_start|>system
+{{ .System }}<|im_end|>
+{{ end }}{{ if .Prompt }}<|im_start|>user
+{{ .Prompt }}<|im_end|>
+{{ end }}<|im_start|>assistant
+{{ end }}{{ .Response }}{{ if .Response }}<|im_end|>{{ end }}

tools/testdata/qwen3.out

@@ -0,0 +1,31 @@
+<|im_start|>system
+You are a knowledgeable assistant. You can answer questions and perform tasks.
+
+# Tools
+
+You may call one or more functions to assist with the user query.
+
+You are provided with function signatures within <tools></tools> XML tags:
+<tools>
+{"type": "function", "function": {"name":"get_current_weather","description":"Get the current weather","parameters":{"type":"object","required":["location","format"],"properties":{"format":{"type":"string","description":"The temperature unit to use. Infer this from the user's location.","enum":["celsius","fahrenheit"]},"location":{"type":"string","description":"The city and state, e.g. San Francisco, CA"}}}}}
+</tools>
+
+For each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:
+<tool_call>
+{"name": <function-name>, "arguments": <args-json-object>}
+</tool_call><|im_end|>
+<|im_start|>user
+What's the weather like today in Paris?<|im_end|>
+<|im_start|>assistant
+<tool_call>
+{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Paris, France"}}
+</tool_call><|im_end|>
+<|im_start|>user
+<tool_response>
+22
+</tool_response><|im_end|>
+<|im_start|>assistant
+The current temperature in Paris, France is 22 degrees Celsius.<|im_end|>
+<|im_start|>user
+What's the weather like today in San Francisco and Toronto?<|im_end|>
+<|im_start|>assistant

tools/tools.go

@@ -0,0 +1,228 @@
+package tools
+
+import (
+	"errors"
+	"io"
+	"log/slog"
+	"strings"
+	gotmpl "text/template"
+
+	jsonv2 "github.com/go-json-experiment/json"
+	jsontext "github.com/go-json-experiment/json/jsontext"
+
+	"github.com/ollama/ollama/api"
+	"github.com/ollama/ollama/template"
+)
+
+// Sentinel errors for parsing states
+var (
+	ErrPartialPrefix = errors.New("partial prefix detected")
+
+	ErrPrefixNotFound = errors.New("prefix not found")
+
+	ErrInvalidToolCall = errors.New("invalid tool call format")
+
+	ErrAccumulateMore = errors.New("need to accumulate more content")
+)
+
+type Parser struct {
+	greedyParse bool
+	prefixFound bool
+	tmpl        gotmpl.Template
+	sb          strings.Builder
+	prefix      string
+	index       int
+	name        string
+	arguments   string
+	Done        bool
+}
+
+// parseJSONToolCalls attempts to parse a JSON string into a slice ToolCalls.
+// It first tries to incrementally decode the JSON to handle partial inputs.
+// Returns:
+//   - []api.ToolCall: The parsed tool calls if successful
+//   - error: ErrPartialJSON if JSON is incomplete, ErrInvalidToolCall if invalid, or nil if successful
+func (p *Parser) parseJSONToolCalls(s string) ([]api.ToolCall, error) {
+	// First try incremental decoding to handle partial JSON
+	dec := jsontext.NewDecoder(strings.NewReader(s))
+	if got, err := dec.ReadValue(); err == nil {
+		s = got.String()
+	}
+
+	// Attempt full unmarshal of the JSON
+	var resp any
+	if err := jsonv2.Unmarshal([]byte(s), &resp); errors.Is(err, io.ErrUnexpectedEOF) {
+		slog.Debug("incomplete JSON detected", "input", s)
+		return nil, ErrAccumulateMore
+	} else if err != nil {
+		slog.Debug("failed to unmarshal response", "error", err)
+		return nil, ErrInvalidToolCall
+	}
+
+	// Collect all nested objects that could contain tool calls
+	objs := collect(resp)
+	if len(objs) == 0 {
+		return nil, ErrInvalidToolCall
+	}
+
+	var toolCalls []api.ToolCall
+	for _, kv := range objs {
+		n, nok := kv[p.name].(string)
+		a, aok := kv[p.arguments].(map[string]any)
+		if nok && aok {
+			toolCalls = append(toolCalls, api.ToolCall{
+				Function: api.ToolCallFunction{
+					Name:      n,
+					Arguments: a,
+				},
+			})
+		}
+	}
+
+	// Valid JSON, no tool calls found
+	if len(toolCalls) == 0 {
+		return nil, ErrInvalidToolCall
+	}
+
+	return toolCalls, nil
+}
+
+// checkPrefix processes a string to find and handle a prefix pattern.
+//
+// Returns:
+//   - The processed string with prefix removed if found
+//   - error: ErrPartialPrefix if prefix is incomplete, ErrPrefixNotFound if not found, or nil if successful
+func (p *Parser) checkPrefix(s string) (string, error) {
+	// Keep original for overlap checks
+	original := s
+	s = strings.TrimSpace(s)
+	if s == "" {
+		return "", nil
+	}
+
+	// If no prefix defined, just return trimmed string
+	if p.prefix == "" {
+		return s, nil
+	}
+
+	// Check for prefix at start of string
+	if processedStr, hasPrefix := strings.CutPrefix(s, p.prefix); hasPrefix {
+		// Found prefix at start - accumulate for potential tool
+		p.prefixFound = true
+		return processedStr, nil
+	}
+
+	// Check if prefix overlaps end of string
+	if overlap := suffixOverlap(original, p.prefix); overlap > 0 {
+		// Return everything except overlapping portion
+		p.sb.Reset()
+		p.sb.WriteString(original[len(original)-overlap:])
+		return original[0 : len(original)-overlap], ErrAccumulateMore
+	}
+
+	// Check if prefix appears in middle of string
+	if idx := strings.Index(original, p.prefix); idx != -1 {
+		// Save remainder starting at prefix for next pass
+		p.sb.Reset()
+		p.sb.WriteString(strings.TrimSpace(original[idx:]))
+		// Return everything before prefix
+		return original[:idx], ErrAccumulateMore
+	}
+
+	// No partial prefix found
+	return s, nil
+}
+
+// Add processes a string input to parse tool calls and content.
+// It handles prefix detection and JSON parsing to extract tool calls.
+//
+// Returns:
+//   - tools: Any parsed tool calls
+//   - content: Non-tool call content
+//   - error: One of the sentinel errors or nil if successful
+func (p *Parser) Add(s string) (tools []api.ToolCall, content string, err error) {
+	p.sb.WriteString(s)
+	s = p.sb.String()
+
+	// Check for prefix pattern in input
+	s, err = p.checkPrefix(s)
+	if err != nil {
+		if s != "" {
+			// Return content before prefix
+			return nil, s, nil
+		}
+		// Need more input to complete prefix
+		return nil, "", ErrAccumulateMore
+	}
+
+	// Exit if prefix exists in template, greedy parsing is off, and prefix not found
+	if !p.greedyParse && !p.prefixFound {
+		p.sb.Reset()
+		return nil, "", ErrPrefixNotFound
+	}
+
+	toolCalls, err := p.parseJSONToolCalls(s)
+	if err != nil {
+		if errors.Is(err, ErrAccumulateMore) {
+			return nil, "", err
+		} else {
+			p.sb.Reset()
+			// Do not try greedy parsing if JSON not found
+			p.greedyParse = false
+			if p.prefix == "" {
+				p.Done = true
+			}
+			if p.prefixFound {
+				// Drop tokens since prefix was found
+				return nil, "", ErrAccumulateMore
+			}
+			return nil, s, nil
+		}
+	}
+
+	for _, tc := range toolCalls {
+		tc.Function.Index = p.index
+		p.index++
+	}
+
+	// Mark as done if no prefix needed
+	if p.prefix == "" {
+		p.Done = true
+	}
+
+	p.sb.Reset()
+	return toolCalls, "", nil
+}
+
+// NewParser creates a new tool call parser from a template. It extracts the tool call format,
+// prefix, and field names from the template to use for parsing tool calls from model output.
+//
+// Returns an error if the template does not contain valid tool call formatting.
+func NewParser(templateToProcess *gotmpl.Template) (Parser, error) {
+	parsed, err := template.Parse(templateToProcess.Root.String())
+	if err != nil {
+		return Parser{}, err
+	}
+
+	tt, err := toolTemplate(parsed)
+	if err != nil {
+		return Parser{}, err
+	}
+
+	tp := toolPrefix(templateToProcess)
+	tp = strings.TrimSpace(tp)
+
+	name, arguments, err := extractToolArgs(tt)
+	if err != nil {
+		return Parser{}, err
+	}
+
+	return Parser{
+		tmpl:        *tt,
+		sb:          strings.Builder{},
+		prefix:      tp,
+		greedyParse: true,
+		name:        name,
+		arguments:   arguments,
+	}, nil
+}

tools/tools_test.go

@@ -0,0 +1,491 @@
+package tools
+
+import (
+	"bytes"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/google/go-cmp/cmp"
+
+	"github.com/ollama/ollama/api"
+	"github.com/ollama/ollama/template"
+)
+
+func readFile(t *testing.T, base, name string) *bytes.Buffer {
+	t.Helper()
+
+	bts, err := os.ReadFile(filepath.Join(base, name))
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	return bytes.NewBuffer(bts)
+}
+
+func TestParseToolCalls(t *testing.T) {
+	p := filepath.Join("testdata")
+	t1 := api.ToolCall{
+		Function: api.ToolCallFunction{
+			Name: "get_current_weather",
+			Arguments: api.ToolCallFunctionArguments{
+				"format":   "fahrenheit",
+				"location": "San Francisco, CA",
+			},
+		},
+	}
+	t2 := api.ToolCall{
+		Function: api.ToolCallFunction{
+			Name: "get_current_weather",
+			Arguments: api.ToolCallFunctionArguments{
+				"format":   "celsius",
+				"location": "Toronto, Canada",
+			},
+		},
+	}
+
+	cases := []struct {
+		name             string
+		model            string
+		output           string
+		expectedToolCall []api.ToolCall
+		expectedTokens   string
+	}{
+		{
+			name:             "mistral malformed json with tool calls prefix",
+			model:            "mistral",
+			output:           `[TOOL_CALLS]  [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_curren}]`,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   "",
+		},
+		{
+			name:             "mistral multiple tool calls without prefix",
+			model:            "mistral",
+			output:           `[{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   "",
+		},
+		{
+			name:  "mistral tool calls with text between no prefix",
+			model: "mistral",
+			output: `[{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}] 
+			model outputs more tokens here and then [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   `model outputs more tokens here and then [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+		},
+		{
+			name:             "mistral valid json with tool calls prefix",
+			model:            "mistral",
+			output:           `[TOOL_CALLS]  [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   "",
+		},
+		{
+			name:  "mistral multiple tool calls with text between and prefix",
+			model: "mistral",
+			output: `[TOOL_CALLS]  [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]
+			model outputs more tokens here and then [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+			expectedToolCall: []api.ToolCall{t1, t2, t1, t2},
+			expectedTokens:   "",
+		},
+		{
+			name:             "mistral incomplete json with tool calls prefix",
+			model:            "mistral",
+			output:           `[TOOL_CALLS]  [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, `,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   "",
+		},
+		{
+			name:  "mistral invalid tool call with explanatory text no prefix",
+			model: "mistral",
+			output: `I'm not aware of that information. However, I can suggest searching for the weather using the "get_current_weather" function:
+
+		[{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   `I'm not aware of that information. However, I can suggest searching for the weather using the "get_current_weather" function: [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+		},
+		{
+			name:             "mistral tool calls without prefix",
+			model:            "mistral",
+			output:           `[{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   "",
+		},
+		{
+			name:  "command r plus tool calls with json block format",
+			model: "command-r-plus",
+			output: "Action: ```json" + `
+		[
+		    {
+		        "tool_name": "get_current_weather",
+		        "parameters": {
+		            "format": "fahrenheit",
+		            "location": "San Francisco, CA"
+		        }
+		    },
+		    {
+		        "tool_name": "get_current_weather",
+		        "parameters": {
+		            "format": "celsius",
+		            "location": "Toronto, Canada"
+		        }
+		    }
+		]
+		` + "```",
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   "",
+		},
+		{
+			name:             "firefunction tool calls with functools prefix",
+			model:            "firefunction",
+			output:           ` functools[{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   "",
+		},
+		{
+			name:  "llama3 groq single tool call with xml tags",
+			model: "llama3-groq-tool-use",
+			output: `<tool_call>
+		{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}
+		</tool_call>`,
+			expectedToolCall: []api.ToolCall{t1},
+			expectedTokens:   "",
+		},
+		{
+			name:             "xlam tool calls with wrapper object",
+			model:            "xlam",
+			output:           `{"tool_calls": [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}},{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]}`,
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   "",
+		},
+		{
+			name:             "qwen2.5-coder single tool call with prefix",
+			model:            "qwen2.5-coder",
+			output:           `<tool_call>{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}</tool_call>`,
+			expectedToolCall: []api.ToolCall{t1},
+			expectedTokens:   "",
+		},
+		{
+			name:             "qwen2.5-coder multiple tool calls with and without prefix",
+			model:            "qwen2.5-coder",
+			output:           `{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} <tool_call>{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}</tool_call> <tool_call>{"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}</tool_call>`,
+			expectedToolCall: []api.ToolCall{t1, t1, t2},
+			expectedTokens:   "",
+		},
+		{
+			name:             "qwen2.5-coder multiple tool calls without prefix",
+			model:            "qwen2.5-coder",
+			output:           `[{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}, {"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   "",
+		},
+		{
+			name:             "qwen2.5-coder plain text response no tool calls",
+			model:            "qwen2.5-coder",
+			output:           "The weather in San Francisco, CA is 70°F and in Toronto, Canada is 20°C.",
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   "The weather in San Francisco, CA is 70°F and in Toronto, Canada is 20°C.",
+		},
+		{
+			name:             "qwen2.5-coder tool calls with trailing text",
+			model:            "qwen2.5-coder",
+			output:           `[{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}, {"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}] some tokens after call`,
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   "some tokens after call",
+		},
+		{
+			name:             "qwen2.5-coder tool calls with initial text",
+			model:            "qwen2.5-coder",
+			output:           `some tokens before call [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}, {"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   `some tokens before call [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}, {"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+		},
+		{
+			name:             "qwen2.5 tool calls with prefix and trailing text",
+			model:            "qwen2.5-coder",
+			output:           `<tool_call> [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}, {"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}] </tool_call> some tokens after call`,
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   "",
+		},
+		{
+			name:             "qwen2.5 tool calls with prefix and initial text",
+			model:            "qwen2.5-coder",
+			output:           `some tokens before call <tool_call> [{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}, {"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}] </tool_call>`,
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   "some tokens before call",
+		},
+		{
+			name:             "qwen2.5 tool calls without prefix and valid tool call",
+			model:            "qwen2.5-coder",
+			output:           `[{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}, {"name": "get_current_weather", "arguments": {"format":"celsius","location":"Toronto, Canada"}}]`,
+			expectedToolCall: []api.ToolCall{t1, t2},
+			expectedTokens:   "",
+		},
+		{
+			name:             "qwen2.5 tool calls without prefix and invalid tool call",
+			model:            "qwen2.5-coder",
+			output:           `[{"options": "foo"}]`,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   `[{"options": "foo"}]`,
+		},
+		{
+			name:             "qwen2.5 tool calls with prefix and invalid tool call",
+			model:            "qwen2.5-coder",
+			output:           `<tool_call> [{"options": "foo"}] </tool_call> `,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   ``,
+		},
+		{
+			name:             "qwen3 tool call with think prefix and tool prefix (sent as a single token)",
+			model:            "qwen3",
+			output:           `<think>Okay, let me think what tool we should use...</think><tool_call>{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}</tool_call>`,
+			expectedToolCall: []api.ToolCall{t1},
+			expectedTokens:   "<think>Okay, let me think what tool we should use...</think>",
+		},
+		{
+			name:             "qwen3 tool call with think prefix, tool prefix, and whitespace (sent as separate tokens)",
+			model:            "qwen3",
+			output:           `<think>Okay, let me think what tool we should use...</think> <tool_call> {"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} </tool_call>`,
+			expectedToolCall: []api.ToolCall{t1},
+			expectedTokens:   "<think>Okay, let me think what tool we should use...</think>",
+		},
+		{
+			name:             "qwen3 empty think prefix without tool prefix and invalid tool call",
+			model:            "qwen3",
+			output:           `<think></think>{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} </tool_call>`,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   `<think></think>{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} </tool_call>`,
+		},
+		{
+			name:             "qwen3 empty think prefix with tool prefix and valid tool call",
+			model:            "qwen3",
+			output:           `<think></think><tool_call>{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}}  </tool_call>`,
+			expectedToolCall: []api.ToolCall{t1},
+			expectedTokens:   `<think></think>`,
+		},
+		{
+			name:             "qwen3 invalid tool call with fake tool prefix (single rune suffix match)",
+			model:            "qwen3",
+			output:           `<think></think>< fakeout{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} </tool_call>`,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   `<think></think>< fakeout{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} </tool_call>`,
+		},
+		{
+			name:             "qwen3 invalid tool call with partial tool prefix (multiple rune suffix match)",
+			model:            "qwen3",
+			output:           `<think></think><tool_c fakeout{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} </tool_call>`,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   `<think></think><tool_c fakeout{"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} </tool_call>`,
+		},
+		{
+			name:             "qwen3 invalid tool call with malformed tool prefix",
+			model:            "qwen3",
+			output:           `<think></think><tool_cfakeout {"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} </tool_call>`,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   `<think></think><tool_cfakeout {"name": "get_current_weather", "arguments": {"format":"fahrenheit","location":"San Francisco, CA"}} </tool_call>`,
+		},
+		{
+			name:             "llama3.2 valid tool call without prefix",
+			model:            "llama3.2",
+			output:           `{"name": "get_current_weather", "parameters": {"format":"fahrenheit","location":"San Francisco, CA"}}`,
+			expectedToolCall: []api.ToolCall{t1},
+			expectedTokens:   "",
+		},
+		{
+			name:             "llama3.2 incomplete tool call without prefix",
+			model:            "llama3.2",
+			output:           `{"name": "get_current_weather", "parameters": {"format":"fahrenheit","location":"San Francisco, `,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   "",
+		},
+		{
+			name:             "llama3.2 tool call with leading text",
+			model:            "llama3.2",
+			output:           `some non json text{"name": "get_current_weather", "parameters": {"format":"fahrenheit","location":"San Francisco, CA"}}`,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   `some non json text{"name": "get_current_weather", "parameters": {"format":"fahrenheit","location":"San Francisco, CA"}}`,
+		},
+		{
+			name:             "llama3.2 tool call with invalid tool prefix (no prefix in template)",
+			model:            "llama3.2",
+			output:           `<tool_call>{"name": "get_current_weather", "parameters": {"format":"fahrenheit","location":"San Francisco, CA"}}`,
+			expectedToolCall: []api.ToolCall{},
+			expectedTokens:   `<tool_call>{"name": "get_current_weather", "parameters": {"format":"fahrenheit","location":"San Francisco, CA"}}`,
+		},
+	}
+
+	var tools []api.Tool
+	if err := json.Unmarshal(readFile(t, p, "tools.json").Bytes(), &tools); err != nil {
+		t.Fatal(err)
+	}
+
+	var messages []api.Message
+	if err := json.Unmarshal(readFile(t, p, "messages.json").Bytes(), &messages); err != nil {
+		t.Fatal(err)
+	}
+
+	for _, tt := range cases {
+		t.Run(tt.name, func(t *testing.T) {
+			tmpl, err := template.Parse(readFile(t, p, fmt.Sprintf("%s.gotmpl", tt.model)).String())
+			if err != nil {
+				t.Fatal(err)
+			}
+
+			t.Run("template", func(t *testing.T) {
+				actual := &bytes.Buffer{} // Create new buffer for each test
+				if err := tmpl.Execute(actual, template.Values{Tools: tools, Messages: messages}); err != nil {
+					t.Fatal(err)
+				}
+
+				if diff := cmp.Diff(actual.String(), readFile(t, p, fmt.Sprintf("%s.out", tt.model)).String()); diff != "" {
+					t.Errorf("mismatch (-got +want):\n%s", diff)
+				}
+			})
+
+			t.Run("parse", func(t *testing.T) {
+				// fmt.Printf("tmpl: %s\n", tmpl.Root.String())
+				tp, err := NewParser(tmpl.Template)
+				if err != nil {
+					t.Fatal(err)
+				}
+				got := []api.ToolCall{}
+				var gotTokens strings.Builder
+
+				var add bool
+				tokens := strings.Fields(tt.output)
+				for _, tok := range tokens {
+					s := " " + tok
+
+					add = true
+					if !tp.Done {
+						toolCalls, content, err := tp.Add(s)
+						if err == nil {
+							if content != "" {
+								fmt.Printf("content: %q\n", content)
+								gotTokens.WriteString(content)
+								add = false
+							} else if len(toolCalls) > 0 {
+								got = append(got, toolCalls...)
+								add = false
+							}
+						} else if errors.Is(err, ErrAccumulateMore) {
+							add = false
+						}
+					}
+					if add {
+						gotTokens.WriteString(s)
+					}
+				}
+
+				// Compare tool calls if we expect any
+				if diff := cmp.Diff(got, tt.expectedToolCall); diff != "" {
+					t.Errorf("tool calls mismatch (-got +want):\n%s", diff)
+				}
+
+				// Compare tokens if we expect any
+				stripped := strings.TrimSpace(gotTokens.String())
+				if diff := cmp.Diff(stripped, tt.expectedTokens); diff != "" {
+					t.Log("actualTokens", stripped, "expectedTokens", tt.expectedTokens)
+					t.Errorf("tokens mismatch (-got +want):\n%s", diff)
+				}
+			})
+		})
+	}
+}
+
+func TestParseJSONToolCalls(t *testing.T) {
+	tests := []struct {
+		name          string
+		input         string
+		parser        *Parser
+		wantToolCalls []api.ToolCall
+		wantErr       error
+	}{
+		{
+			name:   "valid single tool call",
+			input:  `{"name": "test_tool", "arguments": {"arg1": "value1"}}`,
+			parser: &Parser{name: "name", arguments: "arguments"},
+			wantToolCalls: []api.ToolCall{
+				{
+					Function: api.ToolCallFunction{
+						Name: "test_tool",
+						Arguments: map[string]any{
+							"arg1": "value1",
+						},
+					},
+				},
+			},
+			wantErr: nil,
+		},
+		{
+			name:          "incomplete JSON",
+			input:         `{"name": "test_tool", "arguments": {"arg1": `,
+			parser:        &Parser{name: "name", arguments: "arguments"},
+			wantToolCalls: nil,
+			wantErr:       ErrAccumulateMore,
+		},
+		{
+			name:          "invalid JSON",
+			input:         `not json at all`,
+			parser:        &Parser{name: "name", arguments: "arguments"},
+			wantToolCalls: nil,
+			wantErr:       ErrInvalidToolCall,
+		},
+		{
+			name:          "missing required fields",
+			input:         `{"other": "field"}`,
+			parser:        &Parser{name: "name", arguments: "arguments"},
+			wantToolCalls: nil,
+			wantErr:       ErrInvalidToolCall,
+		},
+		{
+			name: "multiple tool calls in array",
+			input: `[
+				{"name": "tool1", "arguments": {"arg1": 1}},
+				{"name": "tool2", "arguments": {"arg2": "value"}}
+			]`,
+			parser: &Parser{name: "name", arguments: "arguments"},
+			wantToolCalls: []api.ToolCall{
+				{
+					Function: api.ToolCallFunction{
+						Name: "tool1",
+						Arguments: map[string]any{
+							"arg1": float64(1),
+						},
+					},
+				},
+				{
+					Function: api.ToolCallFunction{
+						Name: "tool2",
+						Arguments: map[string]any{
+							"arg2": "value",
+						},
+					},
+				},
+			},
+			wantErr: nil,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			gotCalls, err := tt.parser.parseJSONToolCalls(tt.input)
+
+			if err != tt.wantErr {
+				t.Errorf("parseJSONToolCalls() error = %v, want %v", err, tt.wantErr)
+			}
+
+			if len(gotCalls) != 0 && tt.wantErr != nil {
+				t.Errorf("parseJSONToolCalls() valid = %v, want %v", len(gotCalls) == 0, tt.wantErr == nil)
+			}
+
+			if diff := cmp.Diff(gotCalls, tt.wantToolCalls); diff != "" {
+				t.Errorf("parseJSONToolCalls() tool calls mismatch (-got +want):\n%s", diff)
+			}
+		})
+	}
+}

tools/utils.go

@@ -0,0 +1,257 @@
+package tools
+
+import (
+	"bytes"
+	"errors"
+	"log/slog"
+	"slices"
+	"strings"
+	gotmpl "text/template"
+	"text/template/parse"
+
+	jsonv2 "github.com/go-json-experiment/json"
+	"github.com/ollama/ollama/api"
+	"github.com/ollama/ollama/template"
+)
+
+// extractToolCallsFormat traverses a template AST to find text that follows a ".ToolCalls" condition.
+// It walks the template nodes looking for if-statements containing ".ToolCalls" and extracts any
+// immediate text nodes that follow. This is used to identify tool call prefixes and formatting.
+//
+// Returns:
+//   - string: The extracted text following the first ".ToolCalls" condition found
+//   - bool: Whether a ".ToolCalls" condition was found in the template
+func extractToolCallsFormat(tmpl *gotmpl.Template) (string, bool) {
+	if tmpl == nil || tmpl.Tree == nil {
+		slog.Debug("TextAfterToolCalls: template or tree is nil")
+		return "", false
+	}
+
+	var result string
+	var found bool
+
+	var walk func(nodes []parse.Node)
+	walk = func(nodes []parse.Node) {
+		for _, node := range nodes {
+			if found {
+				return
+			}
+
+			switch n := node.(type) {
+			case *parse.IfNode:
+				if isToolCallsNode(n) {
+					// Collect immediate TextNode(s) at start of IfNode's list
+					var sb strings.Builder
+					for _, innerNode := range n.List.Nodes {
+						if tn, ok := innerNode.(*parse.TextNode); ok {
+							sb.Write(tn.Text)
+						} else {
+							// Stop at first non-text node
+							break
+						}
+					}
+					result = sb.String()
+					found = true
+					return
+				}
+				// Recurse into child nodes
+				walk(n.List.Nodes)
+				if n.ElseList != nil {
+					walk(n.ElseList.Nodes)
+				}
+			case *parse.ListNode:
+				walk(n.Nodes)
+			case *parse.RangeNode:
+				walk(n.List.Nodes)
+				if n.ElseList != nil {
+					walk(n.ElseList.Nodes)
+				}
+			case *parse.WithNode:
+				walk(n.List.Nodes)
+				if n.ElseList != nil {
+					walk(n.ElseList.Nodes)
+				}
+			default:
+				// Continue to next node
+				continue
+			}
+
+			if found {
+				return
+			}
+		}
+	}
+
+	walk(tmpl.Tree.Root.Nodes)
+	return result, found
+}
+
+// isToolCallsNode detects if a node's condition includes ".ToolCalls"
+func isToolCallsNode(n *parse.IfNode) bool {
+	for _, cmd := range n.Pipe.Cmds {
+		for _, arg := range cmd.Args {
+			if field, ok := arg.(*parse.FieldNode); ok {
+				if slices.Contains(field.Ident, "ToolCalls") {
+					return true
+				}
+			}
+		}
+	}
+	return false
+}
+
+// TODO(parthsareen): get full prefix from the template instead of just the first token
+
+// toolPrefix returns the prefix for the tool call if it exists from a template
+func toolPrefix(tmpl *gotmpl.Template) string {
+	tokenText, ok := extractToolCallsFormat(tmpl)
+	if !ok {
+		return ""
+	}
+	tokenText = strings.TrimSpace(tokenText)
+	if tokenText == "" {
+		return ""
+	}
+	first := strings.Fields(tokenText)[0]
+
+	start := -1
+	end := -1
+	for i, r := range tokenText {
+		if r == '<' || r == '[' {
+			start = i
+		}
+		if (r == '>' || r == ']') && start != -1 {
+			end = i
+			break
+		}
+	}
+	if start != -1 && end != -1 {
+		// return the token including the [ or < and the ] or >
+		return tokenText[start : end+1]
+	} else if start != -1 {
+		// get until the [ or < - in the case tag was not closed
+		return tokenText[:start]
+	} else if end != -1 {
+		// get after the ] or > - in the case tag was not opened
+		return tokenText[end+1:]
+	}
+	return first
+}
+
+// toolTemplate creates a subtree from the node that ranges over .ToolCalls
+//
+// Returns:
+//   - *gotmpl.Template: The subtree containing the .ToolCalls range
+//   - error: Error if parsing failed
+func toolTemplate(t *template.Template) (*gotmpl.Template, error) {
+	tmpl := t.Subtree(func(n parse.Node) bool {
+		if t, ok := n.(*parse.RangeNode); ok {
+			return slices.Contains(template.Identifiers(t.Pipe), "ToolCalls")
+		}
+
+		return false
+	})
+
+	if tmpl == nil {
+		return nil, errors.New("failed to find tool template")
+	}
+
+	return tmpl, nil
+}
+
+// suffixOverlap returns the length of the longest suffix overlap between two strings
+//
+// Returns:
+//   - int: The length of the longest suffix overlap
+func suffixOverlap(s, prefix string) int {
+	max := min(len(prefix), len(s))
+	for i := max; i > 0; i-- {
+		if strings.HasSuffix(s, prefix[:i]) {
+			return i
+		}
+	}
+	return 0
+}
+
+// extractToolArgs executes a template with a known tool call format to extract the name and arguments
+//
+// Returns:
+//   - string: The name of the tool call
+//   - string: The arguments of the tool call
+//   - error: Error if parsing failed
+func extractToolArgs(tmpl *gotmpl.Template) (name, arguments string, err error) {
+	var b bytes.Buffer
+	if err := tmpl.Execute(&b, map[string][]api.ToolCall{
+		"ToolCalls": {
+			{
+				Function: api.ToolCallFunction{
+					Name: "@@name@@",
+					Arguments: api.ToolCallFunctionArguments{
+						"@@argument@@": 1,
+					},
+				},
+			},
+		},
+	}); err != nil {
+		return "", "", err
+	}
+
+	var obj any
+	err = jsonv2.Unmarshal(b.Bytes(), &obj)
+	if err != nil {
+		return "", "", err
+	}
+
+	var objs []map[string]any
+	switch v := obj.(type) {
+	case map[string]any:
+		objs = []map[string]any{v}
+	case []map[string]any:
+		objs = v
+	case []any:
+		objs = collect(v)
+	}
+	if len(objs) == 0 {
+		return "", "", errors.New("no template objects found")
+	}
+
+	// find the keys that correspond to the name and arguments fields
+	for k, v := range objs[0] {
+		switch v.(type) {
+		case string:
+			name = k
+		case map[string]any:
+			arguments = k
+		}
+	}
+
+	if name == "" || arguments == "" {
+		slog.Debug("missing required fields in tool call template", "name", name, "arguments", arguments)
+		return "", "", errors.New("missing required fields in tool call template")
+	}
+
+	return name, arguments, nil
+}
+
+// collect recursively traverses an object to collect all nested maps
+//
+// Returns:
+//   - []map[string]any: A slice of all nested maps found in the object
+func collect(obj any) []map[string]any {
+	var all []map[string]any
+	switch o := obj.(type) {
+	case map[string]any:
+		all = append(all, o)
+		for _, v := range o {
+			all = append(all, collect(v)...)
+		}
+	case []any:
+		for _, v := range o {
+			all = append(all, collect(v)...)
+		}
+	default:
+		return nil
+	}
+
+	return all
+}

tools/utils_test.go

@@ -0,0 +1,464 @@
+package tools
+
+import (
+	"testing"
+	gotmpl "text/template"
+
+	"github.com/ollama/ollama/template"
+)
+
+func TestExtractToolCallsFormat(t *testing.T) {
+	cases := []struct {
+		name     string
+		template string
+		want     string
+		found    bool
+	}{
+		{
+			name:     "nil template",
+			template: "",
+			want:     "",
+			found:    false,
+		},
+		{
+			name:     "basic tool call with text",
+			template: "{{if .ToolCalls}}Hello world{{end}}",
+			want:     "Hello world",
+			found:    true,
+		},
+		{
+			name:     "tool call with json format",
+			template: "{{if .ToolCalls}}```json\n{{end}}",
+			want:     "```json\n",
+			found:    true,
+		},
+		{
+			name:     "tool call in range",
+			template: "{{range .ToolCalls}}tool: {{.}}{{end}}",
+			want:     "",
+			found:    false,
+		},
+		{
+			name:     "tool call with multiple text nodes",
+			template: "{{if .ToolCalls}}First text{{if .Something}}inner{{end}}Second text{{end}}",
+			want:     "First text",
+			found:    true,
+		},
+		{
+			name:     "nested if without tool calls",
+			template: "{{if .Something}}{{if .OtherThing}}text{{end}}{{end}}",
+			want:     "",
+			found:    false,
+		},
+	}
+
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			tmpl, err := gotmpl.New("test").Parse(tc.template)
+			if err != nil && tc.template != "" {
+				t.Fatalf("failed to parse template: %v", err)
+			}
+
+			got, found := extractToolCallsFormat(tmpl)
+			if got != tc.want {
+				t.Errorf("got text %q, want %q", got, tc.want)
+			}
+			if found != tc.found {
+				t.Errorf("got found %v, want %v", found, tc.found)
+			}
+		})
+	}
+}
+
+func TestToolPrefix(t *testing.T) {
+	cases := []struct {
+		name     string
+		template string
+		want     string
+	}{
+		{
+			name:     "basic tool call with action prefix",
+			template: "{{if .ToolCalls}}Action: ```json{{end}}",
+			want:     "Action:",
+		},
+		{
+			name:     "incomplete functools bracket",
+			template: "{{if .ToolCalls}}functools[{{end}}",
+			want:     "functools",
+		},
+		{
+			name:     "tool call with angle brackets",
+			template: "{{if .ToolCalls}}Hello, world! <tool_call>{{end}}",
+			want:     "<tool_call>",
+		},
+		{
+			name:     "multiple tool call formats",
+			template: "{{if .ToolCalls}}[tool_call] <tool_call>{{end}}",
+			want:     "[tool_call]",
+		},
+		{
+			name:     "single angle bracket tool call",
+			template: "{{if .ToolCalls}}<tool_call>{{end}}",
+			want:     "<tool_call>",
+		},
+		{
+			name:     "incomplete angle bracket after tool call",
+			template: "{{if .ToolCalls}}[tool_call] <{{end}}",
+			want:     "[tool_call]",
+		},
+		{
+			name:     "angle bracket prefix with tool call",
+			template: "{{if .ToolCalls}}> <tool_call>{{end}}",
+			want:     "<tool_call>",
+		},
+		{
+			name:     "uppercase tool call with incomplete bracket",
+			template: "{{if .ToolCalls}}[TOOL_CALL] [{{end}}",
+			want:     "[TOOL_CALL]",
+		},
+		{
+			name:     "uppercase tool call with adjacent bracket",
+			template: "{{if .ToolCalls}}[TOOL_CALL][{{end}}",
+			want:     "[TOOL_CALL]",
+		},
+		{
+			name:     "tool call with pipe delimiters",
+			template: "{{if .ToolCalls}}<|tool_call|>{{end}}",
+			want:     "<|tool_call|>",
+		},
+		{
+			name:     "tool with no prefix",
+			template: "{{if .ToolCalls}}{{end}}",
+			want:     "",
+		},
+	}
+
+	for _, tt := range cases {
+		t.Run(tt.name, func(t *testing.T) {
+			tmpl, err := gotmpl.New("test").Parse(tt.template)
+			if err != nil {
+				t.Fatalf("failed to parse template: %v", err)
+			}
+			got := toolPrefix(tmpl)
+			if got != tt.want {
+				t.Errorf("ToolToken(%q) = %q; want %q", tt.template, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestToolTemplate(t *testing.T) {
+	cases := []struct {
+		name     string
+		template string
+		want     bool
+	}{
+		{
+			name:     "basic tool call range",
+			template: "{{range .ToolCalls}}test{{end}}",
+			want:     true,
+		},
+		{
+			name:     "no tool calls",
+			template: "{{range .Other}}test{{end}}",
+			want:     false,
+		},
+		{
+			name:     "nested tool calls",
+			template: "{{range .Outer}}{{range .ToolCalls}}test{{end}}{{end}}",
+			want:     true,
+		},
+		{
+			name:     "empty template",
+			template: "",
+			want:     false,
+		},
+		{
+			name:     "tool calls in if statement",
+			template: "{{if .ToolCalls}}test{{end}}",
+			want:     false,
+		},
+	}
+
+	for _, tt := range cases {
+		t.Run(tt.name, func(t *testing.T) {
+			tmpl, err := gotmpl.New("test").Parse(tt.template)
+			if err != nil {
+				t.Fatalf("failed to parse template: %v", err)
+			}
+
+			parsed, err := template.Parse(tmpl.Root.String())
+			if err != nil {
+				t.Fatalf("failed to parse template: %v", err)
+			}
+
+			_, err = toolTemplate(parsed)
+			if err != nil && tt.want {
+				t.Errorf("toolTemplate() = %v; want %v", err, tt.want)
+			}
+		})
+	}
+}
+
+func TestSuffixOverlap(t *testing.T) {
+	cases := []struct {
+		name string
+		s    string
+		d    string
+		want int
+	}{
+		{
+			name: "no overlap",
+			s:    "hello world",
+			d:    "",
+			want: 0,
+		},
+		{
+			name: "full overlap",
+			s:    "<tool_call>",
+			d:    "<tool_call>",
+			want: 11,
+		},
+		{
+			name: "partial overlap",
+			s:    "text <tool_call>",
+			d:    "<tool_call>",
+			want: 11,
+		},
+		{
+			name: "delimiter longer than string",
+			s:    "<tool>",
+			d:    "<tool_call>",
+			want: 0,
+		},
+		{
+			name: "empty string",
+			s:    "",
+			d:    "<tool_call>",
+			want: 0,
+		},
+		{
+			name: "empty delimiter",
+			s:    "<tool_call>",
+			d:    "",
+			want: 0,
+		},
+		{
+			name: "single char overlap",
+			s:    "test<",
+			d:    "<tool_call>",
+			want: 1,
+		},
+		{
+			name: "partial tool call",
+			s:    "hello <tool_",
+			d:    "<tool_call>",
+			want: 6,
+		},
+	}
+
+	for _, tt := range cases {
+		t.Run(tt.name, func(t *testing.T) {
+			got := suffixOverlap(tt.s, tt.d)
+			if got != tt.want {
+				t.Errorf("suffixOverlap(%q, %q) = %d; want %d", tt.s, tt.d, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestExtractToolArgs(t *testing.T) {
+	cases := []struct {
+		name     string
+		template string
+		want     string
+		ok       bool
+	}{
+		{
+			name:     "basic tool call with text after",
+			template: `{{if .ToolCalls}}tool response{{end}}`,
+			want:     "tool response",
+			ok:       true,
+		},
+		{
+			name:     "tool call with mixed content after",
+			template: `{{if .ToolCalls}}<tool_call>{{.Something}}{{end}}`,
+			want:     "<tool_call>",
+			ok:       true,
+		},
+		{
+			name:     "tool call with no text after",
+			template: `{{if .ToolCalls}}{{.Something}}{{end}}`,
+			want:     "",
+			ok:       true,
+		},
+		{
+			name:     "nested tool call",
+			template: `{{if .Something}}{{if .ToolCalls}}[TOOL_CALL]{{end}}{{end}}`,
+			want:     "[TOOL_CALL]",
+			ok:       true,
+		},
+		{
+			name:     "no tool calls",
+			template: `{{if .Something}}no tools here{{end}}`,
+			want:     "",
+			ok:       false,
+		},
+		{
+			name:     "empty template",
+			template: ``,
+			want:     "",
+			ok:       false,
+		},
+		{
+			name:     "multiple tool calls sections",
+			template: `{{if .ToolCalls}}first{{end}}{{if .ToolCalls}}second{{end}}`,
+			want:     "first",
+			ok:       true,
+		},
+		{
+			name:     "range over tool calls",
+			template: `{{if .ToolCalls}}{{range .ToolCalls}}tool{{end}}{{end}}`,
+			want:     "",
+			ok:       true,
+		},
+		{
+			name:     "tool calls with pipe delimiters",
+			template: `{{if .ToolCalls}}<|tool|>{{end}}`,
+			want:     "<|tool|>",
+			ok:       true,
+		},
+		{
+			name:     "tool calls with nested template",
+			template: `{{if .ToolCalls}}{{template "tool" .}}{{end}}`,
+			want:     "",
+			ok:       true,
+		},
+		{
+			name:     "tool calls with whitespace variations",
+			template: `{{if .ToolCalls}}  tool  {{end}}`,
+			want:     "  tool  ",
+			ok:       true,
+		},
+	}
+
+	for _, tt := range cases {
+		t.Run(tt.name, func(t *testing.T) {
+			tmpl, err := gotmpl.New("test").Parse(tt.template)
+			if err != nil {
+				t.Fatalf("failed to parse template: %v", err)
+			}
+
+			got, ok := extractToolCallsFormat(tmpl)
+			if got != tt.want {
+				t.Errorf("TextAfterToolCalls() got = %q, want %q", got, tt.want)
+			}
+			if ok != tt.ok {
+				t.Errorf("TextAfterToolCalls() ok = %v, want %v", ok, tt.ok)
+			}
+		})
+	}
+}
+
+func TestCollect(t *testing.T) {
+	cases := []struct {
+		name string
+		obj  any
+		want []map[string]any
+	}{
+		{
+			name: "simple map",
+			obj: map[string]any{
+				"key": "value",
+			},
+			want: []map[string]any{
+				{"key": "value"},
+			},
+		},
+		{
+			name: "nested map",
+			obj: map[string]any{
+				"outer": map[string]any{
+					"inner": "value",
+				},
+			},
+			want: []map[string]any{
+				{"outer": map[string]any{"inner": "value"}},
+				{"inner": "value"},
+			},
+		},
+		{
+			name: "array of maps",
+			obj: []any{
+				map[string]any{"key1": "val1"},
+				map[string]any{"key2": "val2"},
+			},
+			want: []map[string]any{
+				{"key1": "val1"},
+				{"key2": "val2"},
+			},
+		},
+		{
+			name: "deeply nested",
+			obj: map[string]any{
+				"l1": map[string]any{
+					"l2": map[string]any{
+						"l3": "value",
+					},
+				},
+			},
+			want: []map[string]any{
+				{"l1": map[string]any{"l2": map[string]any{"l3": "value"}}},
+				{"l2": map[string]any{"l3": "value"}},
+				{"l3": "value"},
+			},
+		},
+		{
+			name: "non-map value",
+			obj:  "string",
+			want: nil,
+		},
+	}
+
+	for _, tt := range cases {
+		t.Run(tt.name, func(t *testing.T) {
+			got := collect(tt.obj)
+			if len(got) != len(tt.want) {
+				t.Errorf("collect() got %d maps, want %d", len(got), len(tt.want))
+				return
+			}
+
+			// Compare each map in the result
+			for i := range tt.want {
+				if !mapsEqual(got[i], tt.want[i]) {
+					t.Errorf("collect() map[%d] = %v, want %v", i, got[i], tt.want[i])
+				}
+			}
+		})
+	}
+}
+
+// mapsEqual compares two maps for deep equality
+func mapsEqual(m1, m2 map[string]any) bool {
+	if len(m1) != len(m2) {
+		return false
+	}
+	for k, v1 := range m1 {
+		v2, ok := m2[k]
+		if !ok {
+			return false
+		}
+		switch val1 := v1.(type) {
+		case map[string]any:
+			val2, ok := v2.(map[string]any)
+			if !ok || !mapsEqual(val1, val2) {
+				return false
+			}
+		default:
+			if v1 != v2 {
+				return false
+			}
+		}
+	}
+	return true
+}