CERBERUS/ollama
1
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-08-27 04:06:17 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 27

scheduler: default parallel=1 for qwen3next/lfm (#14103)

Browse source
This commit is contained in:
Jeffrey Morgan 2026-02-05 12:48:25 -08:00 • committed by GitHub
parent c61023f554
commit e36f389e82
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
server/sched.go
View file

@ -417,9 +417,9 @@ func (s *Scheduler) load(req *LlmRequest, f *ggml.GGML, systemInfo ml.SystemInfo
numParallel = 1
}
// `mllama`, `qwen3vl`, and `qwen3vlmoe` are snowflakes and uses an encoder cache which cannot be used with num_parallel > 1
// Some architectures are not safe with num_parallel > 1.
// ref: https://github.com/ollama/ollama/issues/4165
if slices.Contains([]string{"mllama", "qwen3vl", "qwen3vlmoe"}, req.model.Config.ModelFamily) && numParallel != 1 {
if slices.Contains([]string{"mllama", "qwen3vl", "qwen3vlmoe", "qwen3next", "lfm2", "lfm2moe"}, req.model.Config.ModelFamily) && numParallel != 1 {
numParallel = 1
slog.Warn("model architecture does not currently support parallel requests", "architecture", req.model.Config.ModelFamily)
}