baalajimaestro/ollama
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

runner.go: Use correct index when retrieving embedding results

Browse source 
This doesn't have any impact currently because NUM_PARALLEL is forced
to 1 for embeddings, so both indicies will always be 0.
This commit is contained in:
Jesse Gross 2024-11-19 16:41:42 -08:00 committed by Jesse Gross
parent ecf41eed05
commit 5f68fcab12
1 changed files with 1 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
llama/runner/runner.go
View file

@ -454,7 +454,7 @@ func (s *Server) processBatch(tokenBatch *llama.Batch, embedBatch *llama.Batch)
// if done processing the prompt, generate an embedding and return // if done processing the prompt, generate an embedding and return
if seq.embeddingOnly { if seq.embeddingOnly {
embed := s.lc.GetEmbeddingsSeq(i) embed := s.lc.GetEmbeddingsSeq(seq.cache.Id)
if embed == nil { if embed == nil {
embed = s.lc.GetEmbeddingsIth(seq.iBatch) embed = s.lc.GetEmbeddingsIth(seq.iBatch)
} }