💡 All-in-one AI framework for semantic search, LLM orchestration and language model workflows
Python agentsaiai-agentsembeddingsinformation-retrievallanguage-modellarge-language-modelsllmnlppythonragretrieval-augmented-generationsearchsearch-enginesemantic-searchsentence-embeddingstransformerstxtaivector-databasevector-search
Only pass tokenizer to pipeline when necessary, closes #363
davidmezzettia8b1a7b4b59f
diff --git a/src/python/txtai/pipeline/hfpipeline.py b/src/python/txtai/pipeline/hfpipeline.pyindex 88a59df0..171a8fec 100644--- a/src/python/txtai/pipeline/hfpipeline.py+++ b/src/python/txtai/pipeline/hfpipeline.py@@ -41,7 +41,7 @@ class HFPipeline(Tensors):self.pipeline = pipeline(task, model=Models.load(path[0], config, task), tokenizer=path[1], device=deviceid)else:- self.pipeline = pipeline(task, model=path, tokenizer=path, device=deviceid)+ self.pipeline = pipeline(task, model=path, device=deviceid)# Model quantization. Compresses model to int8 precision, improves runtime performance. Only supported on CPU.if deviceid == -1 and quantize: