Merge pull request microsoft#1 from microsoft/dolly-hf

Updated to point to Dolly 2 optimized version on HuggingFace
byte-rose · May 23, 2023 · 230e156 · 230e156
2 parents ebb8057 + ee83f11
commit 230e156
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 3 deletions.
diff --git a/PodcastSocialMediaCopilot.py b/PodcastSocialMediaCopilot.py
@@ -102,8 +102,9 @@
 
 # Step 2 - Make a call to a local Dolly 2.0 model optimized for Windows to extract the name of who I'm interviewing from the transcript
 print("Calling a local Dolly 2.0 model optimized for Windows to extract the name of the podcast guest...\n")
-tokenizer = AutoTokenizer.from_pretrained("models/dolly_v2_7b_merged_fp16.0.0.1", padding_side="left")
-model = ORTModelForCausalLM.from_pretrained("models/dolly_v2_7b_merged_fp16.0.0.1", provider="DmlExecutionProvider", use_cache=True, use_merged=True, use_io_binding=False)
+repo_id = "microsoft/dolly-v2-7b-olive-optimized"
+tokenizer = AutoTokenizer.from_pretrained(repo_id, padding_side="left")
+model = ORTModelForCausalLM.from_pretrained(repo_id, provider="DmlExecutionProvider", use_cache=True, use_merged=True, use_io_binding=False)
 streamer = TextStreamer(tokenizer, skip_prompt=True)
 generate_text = InstructionTextGenerationPipeline(model=model, streamer=streamer, tokenizer=tokenizer, max_new_tokens=128, return_full_text=True, task="text-generation")
 hf_pipeline = HuggingFacePipeline(pipeline=generate_text)

diff --git a/requirements.txt b/requirements.txt
@@ -2,7 +2,6 @@ torch
 transformers
 optimum
 onnx
-wheels/dolly_v2_wheel.0.0.4/onnxruntime_directml-1.16.0-cp39-cp39-win_amd64.whl
 openai-whisper
 langchain
 pydub