Compare commits

..

20 Commits

Author SHA1 Message Date
Daniel Hiltgen
b662e4706e Remove default auto from help message
This may confuse users thinking "auto" is an acceptable string - it must be numeric
2024-07-01 16:01:01 -07:00
Daniel Hiltgen
be31611ff1 Fix case for NumCtx 2024-07-01 16:01:01 -07:00
Daniel Hiltgen
02ba11b614 Document concurrent behavior and settings 2024-07-01 16:01:01 -07:00
Daniel Hiltgen
03bb60e036 Sort the ps output
Provide consistent ordering for the ps command - longest duration listed first
2024-07-01 16:01:01 -07:00
Daniel Hiltgen
976fc86978 Disable concurrency for AMD + Windows
Until ROCm v6.2 ships, we wont be able to get accurate free memory
reporting on windows, which makes automatic concurrency too risky.
Users can still opt-in but will need to pay attention to model sizes otherwise they may thrash/page VRAM or cause OOM crashes.
All other platforms and GPUs have accurate VRAM reporting wired
up now, so we can turn on concurrency by default.
2024-07-01 16:01:01 -07:00
Daniel Hiltgen
9bceb3b55e Enable concurrency by default
This adjusts our default settings to enable multiple models and parallel
requests to a single model.  Users can still override these by the same
env var settings as before.  Parallel has a direct impact on
num_ctx, which in turn can have a significant impact on small VRAM GPUs
so this change also refines the algorithm so that when parallel is not
explicitly set by the user, we try to find a reasonable default that fits
the model on their GPU(s).  As before, multiple models will only load
concurrently if they fully fit in VRAM.
2024-07-01 16:01:01 -07:00
RAPID ARCHITECT
7add3e5267 Update README.md (#5214)
* Update README.md

Added Mesop example to web & desktop

* Update README.md

---------

Co-authored-by: Jeffrey Morgan <jmorganca@gmail.com>
2024-07-01 16:01:01 -07:00
Eduard
c4f2236cf9 Update gpu.md (#5382)
Runs fine on a NVIDIA GeForce GTX 1050 Ti
2024-07-01 16:01:01 -07:00
Jeffrey Morgan
b7ccdcef94 Update api.md 2024-07-01 16:01:01 -07:00
Jeffrey Morgan
1f4f46800c Do not shift context for sliding window models (#5368)
* Do not shift context for sliding window models

* truncate prompt > 2/3 tokens

* only target gemma2
2024-07-01 16:01:01 -07:00
royjhan
42574d3b11 Include Show Info in Interactive (#5342) 2024-07-01 16:01:01 -07:00
royjhan
7bd7e113e3 Ollama Show: Check for Projector Type (#5307)
* Check exists projtype

* Maintain Ordering
2024-07-01 16:01:01 -07:00
royjhan
20240927f8 Update docs (#5312) 2024-07-01 16:01:01 -07:00
Michael Yang
3af1c58146 gemma2 graph 2024-07-01 16:01:01 -07:00
Michael
d90b27a57f update readme for gemma 2 (#5333)
* update readme for gemma 2
2024-07-01 16:01:01 -07:00
Michael Yang
b7ce14c764 zip: prevent extracting files into parent dirs (#5314) 2024-07-01 16:01:01 -07:00
Jeffrey Morgan
161229a153 llm: architecture patch (#5316) 2024-07-01 16:01:01 -07:00
Josh Yan
bd8d680e26 refactor error 2024-07-01 15:57:57 -07:00
Josh Yan
a562b9069f refactor error 2024-07-01 15:56:47 -07:00
Josh Yan
5d76e78c2f add error message for unsupported arch 2024-07-01 15:43:03 -07:00
4 changed files with 7 additions and 82 deletions

View File

@@ -70,12 +70,12 @@ RUN OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu_avx" sh gen_linux.sh
FROM --platform=linux/amd64 cpu-builder-amd64 AS cpu_avx2-build-amd64
RUN OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu_avx2" sh gen_linux.sh
FROM --platform=linux/arm64 rockylinux:8 AS cpu-builder-arm64
FROM --platform=linux/arm64 centos:7 AS cpu-builder-arm64
ARG CMAKE_VERSION
ARG GOLANG_VERSION
COPY ./scripts/rh_linux_deps.sh /
RUN CMAKE_VERSION=${CMAKE_VERSION} GOLANG_VERSION=${GOLANG_VERSION} sh /rh_linux_deps.sh
ENV PATH /opt/rh/gcc-toolset-10/root/usr/bin:$PATH
ENV PATH /opt/rh/devtoolset-10/root/usr/bin:$PATH
COPY --from=llm-code / /go/src/github.com/ollama/ollama/
ARG OLLAMA_CUSTOM_CPU_DEFS
ARG CGO_CFLAGS

View File

@@ -124,7 +124,7 @@ func ParseFile(r io.Reader) (*File, error) {
case stateComment, stateNil:
// pass
case stateValue:
s, ok := unquote(strings.TrimSpace(b.String()))
s, ok := unquote(b.String())
if !ok || isSpace(r) {
if _, err := b.WriteRune(r); err != nil {
return nil, err
@@ -158,7 +158,7 @@ func ParseFile(r io.Reader) (*File, error) {
case stateComment, stateNil:
// pass; nothing to flush
case stateValue:
s, ok := unquote(strings.TrimSpace(b.String()))
s, ok := unquote(b.String())
if !ok {
return nil, io.ErrUnexpectedEOF
}

View File

@@ -22,13 +22,7 @@ ADAPTER adapter1
LICENSE MIT
PARAMETER param1 value1
PARAMETER param2 value2
TEMPLATE """{{ if .System }}<|start_header_id|>system<|end_header_id|>
{{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|>
{{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|>
{{ .Response }}<|eot_id|>"""
TEMPLATE template1
`
reader := strings.NewReader(input)
@@ -42,40 +36,7 @@ TEMPLATE """{{ if .System }}<|start_header_id|>system<|end_header_id|>
{Name: "license", Args: "MIT"},
{Name: "param1", Args: "value1"},
{Name: "param2", Args: "value2"},
{Name: "template", Args: "{{ if .System }}<|start_header_id|>system<|end_header_id|>\n\n{{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|>\n\n{{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|>\n\n{{ .Response }}<|eot_id|>"},
}
assert.Equal(t, expectedCommands, modelfile.Commands)
}
func TestParseFileTrimSpace(t *testing.T) {
input := `
FROM " model 1"
ADAPTER adapter3
LICENSE "MIT "
PARAMETER param1 value1
PARAMETER param2 value2
TEMPLATE """ {{ if .System }}<|start_header_id|>system<|end_header_id|>
{{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|>
{{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|>
{{ .Response }}<|eot_id|> """
`
reader := strings.NewReader(input)
modelfile, err := ParseFile(reader)
require.NoError(t, err)
expectedCommands := []Command{
{Name: "model", Args: " model 1"},
{Name: "adapter", Args: "adapter3"},
{Name: "license", Args: "MIT "},
{Name: "param1", Args: "value1"},
{Name: "param2", Args: "value2"},
{Name: "template", Args: " {{ if .System }}<|start_header_id|>system<|end_header_id|>\n\n{{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|>\n\n{{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|>\n\n{{ .Response }}<|eot_id|> "},
{Name: "template", Args: "template1"},
}
assert.Equal(t, expectedCommands, modelfile.Commands)
@@ -87,26 +48,6 @@ func TestParseFileFrom(t *testing.T) {
expected []Command
err error
}{
{
"FROM \"FOO BAR \"",
[]Command{{Name: "model", Args: "FOO BAR "}},
nil,
},
{
"FROM \"FOO BAR\"\nPARAMETER param1 value1",
[]Command{{Name: "model", Args: "FOO BAR"}, {Name: "param1", Args: "value1"}},
nil,
},
{
"FROM FOOO BAR ",
[]Command{{Name: "model", Args: "FOOO BAR"}},
nil,
},
{
"FROM /what/is/the path ",
[]Command{{Name: "model", Args: "/what/is/the path"}},
nil,
},
{
"FROM foo",
[]Command{{Name: "model", Args: "foo"}},
@@ -145,11 +86,6 @@ func TestParseFileFrom(t *testing.T) {
[]Command{{Name: "param1", Args: "value1"}, {Name: "model", Args: "foo"}},
nil,
},
{
"PARAMETER what the \nFROM lemons make lemonade ",
[]Command{{Name: "what", Args: "the"}, {Name: "model", Args: "lemons make lemonade"}},
nil,
},
}
for _, c := range cases {
@@ -463,7 +399,7 @@ func TestParseFileParameters(t *testing.T) {
"mirostat_eta 1.0": {"mirostat_eta", "1.0"},
"penalize_newline true": {"penalize_newline", "true"},
"stop ### User:": {"stop", "### User:"},
"stop ### User: ": {"stop", "### User:"},
"stop ### User: ": {"stop", "### User: "},
"stop \"### User:\"": {"stop", "### User:"},
"stop \"### User: \"": {"stop", "### User: "},
"stop \"\"\"### User:\"\"\"": {"stop", "### User:"},

View File

@@ -6,21 +6,10 @@ set -ex
MACHINE=$(uname -m)
if grep -i "centos" /etc/system-release >/dev/null; then
# As of 7/1/2024 mirrorlist.centos.org has been taken offline, so adjust accordingly
sed -i s/mirror.centos.org/vault.centos.org/g /etc/yum.repos.d/*.repo
sed -i s/^#.*baseurl=http/baseurl=http/g /etc/yum.repos.d/*.repo
sed -i s/^mirrorlist=http/#mirrorlist=http/g /etc/yum.repos.d/*.repo
# Centos 7 derivatives have too old of a git version to run our generate script
# uninstall and ignore failures
yum remove -y git
yum -y install epel-release centos-release-scl
# The release packages reinstate the mirrors, undo that again
sed -i s/mirror.centos.org/vault.centos.org/g /etc/yum.repos.d/*.repo
sed -i s/^#.*baseurl=http/baseurl=http/g /etc/yum.repos.d/*.repo
sed -i s/^mirrorlist=http/#mirrorlist=http/g /etc/yum.repos.d/*.repo
yum -y install dnf
if [ "${MACHINE}" = "x86_64" ]; then
yum -y install https://repo.ius.io/ius-release-el7.rpm