Only load supported models on new engine (#11362)

* Only load supported models on new engine

Verify the model is supported before trying to load

* int: testcase for all library models
This commit is contained in:
Daniel Hiltgen
2025-07-11 12:21:54 -07:00
committed by GitHub
parent 35fda7b4af
commit f8a6e88819
4 changed files with 261 additions and 0 deletions

View File

@@ -2,6 +2,7 @@ package llama
import (
"cmp"
"fmt"
"math"
"github.com/ollama/ollama/fs"
@@ -33,6 +34,14 @@ type Model struct {
}
func New(c fs.Config) (model.Model, error) {
// This model currently only supports the gpt2 tokenizer
if c.String("tokenizer.ggml.model") == "llama" {
return nil, fmt.Errorf("unsupported tokenizer: llama")
}
// Best effort detection of library/deepseek-coder model(s) which are incompatible
if c.String("general.name") == "deepseek-ai" {
return nil, fmt.Errorf("unsupported model: %s", c.String("general.name"))
}
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}{1,3}| ?[^\s\p{L}\p{N}]+[\r\n]*|\s*[\r\n]+|\s+(?!\S)|\s+`),

View File

@@ -2,7 +2,9 @@ package qwen2
import (
"cmp"
"fmt"
"math"
"strings"
"github.com/ollama/ollama/fs"
"github.com/ollama/ollama/kvcache"
@@ -126,6 +128,14 @@ func (m Model) Shift(ctx ml.Context, layer int, key, shift ml.Tensor) (ml.Tensor
}
func New(c fs.Config) (model.Model, error) {
// This model currently only supports the gpt2 tokenizer
if c.String("tokenizer.ggml.model") == "llama" {
return nil, fmt.Errorf("unsupported tokenizer: llama")
}
// detect library/qwen model(s) which are incompatible
if strings.HasPrefix(c.String("general.name"), "Qwen2-beta") {
return nil, fmt.Errorf("unsupported model: %s", c.String("general.name"))
}
m := Model{
Layers: make([]DecoderLayer, c.Uint("block_count")),
BytePairEncoding: model.NewBytePairEncoding(