pali112/ollama
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

staticcheck

Browse Source
This commit is contained in:
Michael Yang
2025-11-16 19:09:23 -08:00
parent 78a75a30d8
commit d3228355be
44 changed files with 153 additions and 156 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
model/models/gemma2/model.go
View File

@@ -128,7 +128,7 @@ func (sa *SelfAttention) Forward(ctx ml.Context, hiddenState, positionIDs ml.Ten
}
func (m *Model) Shift(ctx ml.Context, layer int, key, shift ml.Tensor) (ml.Tensor, error) {
return fast.RoPE(ctx, key, shift, m.Options.attnKeyLen, m.Options.ropeBase, 1/m.Options.ropeScale, rope.WithTypeNeoX()), nil
return fast.RoPE(ctx, key, shift, m.attnKeyLen, m.ropeBase, 1/m.ropeScale, rope.WithTypeNeoX()), nil
}
type MLP struct {
@@ -178,10 +178,10 @@ func (m *Model) Forward(ctx ml.Context, batch input.Batch) (ml.Tensor, error) {
positions := ctx.Input().FromInts(batch.Positions, len(batch.Positions))
hiddenState := m.TokenEmbedding.Forward(ctx, batch.Inputs)
hiddenState = hiddenState.Scale(ctx, math.Sqrt(float64(m.Options.hiddenSize)))
hiddenState = hiddenState.Scale(ctx, math.Sqrt(float64(m.hiddenSize)))
if len(m.Layers) == gemma27BLayerCount {
m.Options.largeModelScaling = true
m.largeModelScaling = true
}
for i, layer := range m.Layers {
@@ -202,9 +202,9 @@ func (m *Model) Forward(ctx ml.Context, batch input.Batch) (ml.Tensor, error) {
hiddenState = m.Output.Forward(ctx, hiddenState)
// final logit softcap
hiddenState = hiddenState.Scale(ctx, 1.0/float64(m.Options.finalLogitSoftcap))
hiddenState = hiddenState.Scale(ctx, 1.0/float64(m.finalLogitSoftcap))
hiddenState = hiddenState.Tanh(ctx)
return hiddenState.Scale(ctx, float64(m.Options.finalLogitSoftcap)), nil
return hiddenState.Scale(ctx, float64(m.finalLogitSoftcap)), nil
}
func init() {

8
model/models/gemma3/model.go
View File

@@ -96,15 +96,15 @@ func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input
return nil, err
}
f32s, err := m.ImageProcessor.ProcessImage(image)
f32s, err := m.ProcessImage(image)
if err != nil {
return nil, err
}
pixelValues := ctx.Input().FromFloats(f32s,
m.ImageProcessor.imageSize,
m.ImageProcessor.imageSize,
m.ImageProcessor.numChannels,
m.imageSize,
m.imageSize,
m.numChannels,
)
visionOutputs := m.VisionModel.Forward(ctx, pixelValues)

8
model/models/gemma3/model_text.go
View File

@@ -111,12 +111,12 @@ func (sa *TextSelfAttention) Forward(ctx ml.Context, layer int, hiddenState, pos
}
func (m *TextModel) Shift(ctx ml.Context, layer int, key, shift ml.Tensor) (ml.Tensor, error) {
ropeBase := m.TextConfig.ropeLocalBase
ropeBase := m.ropeLocalBase
if (layer+1)%gemmaGlobalCacheCount == 0 {
ropeBase = m.TextConfig.ropeGlobalBase
ropeBase = m.ropeGlobalBase
}
return fast.RoPE(ctx, key, shift, m.TextConfig.attnKeyLen, ropeBase, 1/m.TextConfig.ropeScale, rope.WithTypeNeoX()), nil
return fast.RoPE(ctx, key, shift, m.attnKeyLen, ropeBase, 1/m.ropeScale, rope.WithTypeNeoX()), nil
}
type TextMLP struct {
@@ -166,7 +166,7 @@ func (m *TextModel) Forward(ctx ml.Context, batch input.Batch, cache kvcache.Cac
positions := ctx.Input().FromInts(batch.Positions, len(batch.Positions))
hiddenState := m.TokenEmbedding.Forward(ctx, batch.Inputs)
hiddenState = hiddenState.Scale(ctx, math.Sqrt(float64(m.TextConfig.hiddenSize)))
hiddenState = hiddenState.Scale(ctx, math.Sqrt(float64(m.hiddenSize)))
// set image embeddings
var except []int

6
model/models/mistral3/model.go
View File

@@ -53,7 +53,7 @@ func New(c fs.Config) (model.Model, error) {
MultiModalProjector: newMultiModalProjector(c),
}
m.Cache = kvcache.NewCausalCache(m.TextModel.Shift)
m.Cache = kvcache.NewCausalCache(m.Shift)
return m, nil
}
@@ -109,12 +109,12 @@ func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input
return nil, err
}
f32s, size, err := m.ImageProcessor.ProcessImage(image)
f32s, size, err := m.ProcessImage(image)
if err != nil {
return nil, err
}
pixelValues := ctx.Input().FromFloats(f32s, size.X, size.Y, m.ImageProcessor.numChannels)
pixelValues := ctx.Input().FromFloats(f32s, size.X, size.Y, m.numChannels)
visionOutputs := m.VisionModel.Forward(ctx, pixelValues)
features, size := m.MultiModalProjector.Forward(ctx, visionOutputs, size)

2
model/models/mistral3/model_vision.go
View File

@@ -133,7 +133,7 @@ func (m *VisionModel) Forward(ctx ml.Context, pixelValues ml.Tensor) ml.Tensor {
hiddenStates := m.PatchEmbedding.Forward(ctx, pixelValues, m.patchSize, m.patchSize, 0, 0, 1, 1)
hiddenStates = hiddenStates.Reshape(ctx, numPatches, m.hiddenSize)
hiddenStates = hiddenStates.Permute(ctx, 1, 0, 2, 3).Contiguous(ctx)
hiddenStates = m.EncoderNorm.Forward(ctx, hiddenStates, m.VisionModelOptions.eps)
hiddenStates = m.EncoderNorm.Forward(ctx, hiddenStates, m.eps)
// Prepare position IDs for 2D rope
positions := make([]int32, numPatches)

4
model/models/mllama/model.go
View File

@@ -54,7 +54,7 @@ func New(c fs.Config) (model.Model, error) {
encoderCache := kvcache.NewEncoderCache()
encoderCache.SetConfig(ml.CacheConfig{})
m.Cache = kvcache.NewWrapperCache(encoderCache, kvcache.NewCausalCache(m.TextModel.Shift))
m.Cache = kvcache.NewWrapperCache(encoderCache, kvcache.NewCausalCache(m.Shift))
return &m, nil
}
@@ -69,7 +69,7 @@ func (m *Model) EncodeMultimodal(ctx ml.Context, multimodalData []byte) ([]input
return nil, err
}
f32s, ratio, err := m.ImageProcessor.ProcessImage(image)
f32s, ratio, err := m.ProcessImage(image)
if err != nil {
return nil, err
}

7
model/models/qwen25vl/model.go
View File

@@ -48,7 +48,7 @@ func New(c fs.Config) (model.Model, error) {
ImageProcessor: newImageProcessor(c),
}
m.Cache = kvcache.NewCausalCache(m.TextModel.Shift)
m.Cache = kvcache.NewCausalCache(m.Shift)
return m, nil
}
@@ -59,14 +59,13 @@ func (m *Model) PixelValues(ctx ml.Context, multimodalData []byte) (ml.Tensor, *
return nil, nil, err
}
f32s, grid, err := m.ImageProcessor.ProcessImage(image)
f32s, grid, err := m.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
patchDim := m.numChannels * m.temporalPatchSize * m.patchSize * m.patchSize
numPatches := grid.Temporal * grid.Height * grid.Width
pixelValues := ctx.Input().FromFloats(f32s, patchDim, numPatches)

2
model/models/qwen25vl/model_vision.go
View File

@@ -228,7 +228,7 @@ func (m *VisionModel) Forward(ctx ml.Context, pixelValues ml.Tensor, grid *Grid)
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)
mask := blockDiagonalMask(ctx, hiddenStates.Dim(1), bounds, m.numHeads)
// Apply encoder layers
for i, layer := range m.Layers {
if slices.Contains(m.fullAttnBlocks, int32(i)) {

2
model/models/qwen3/model.go
View File

@@ -203,7 +203,7 @@ func (m *Model) forward(ctx ml.Context, batch input.Batch) (ml.Tensor, error) {
}
func (m *Model) Shift(ctx ml.Context, layer int, key, shift ml.Tensor) (ml.Tensor, error) {
return m.Options.applyRotaryPositionEmbeddings(ctx, key, shift), nil
return m.applyRotaryPositionEmbeddings(ctx, key, shift), nil
}
var _ model.Model = (*Model)(nil)