kenlkehl commited on
Commit
d9d81e4
Β·
verified Β·
1 Parent(s): c4d8a31

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +3 -3
app.py CHANGED
@@ -276,7 +276,7 @@ def load_tagger_model(model_path: str) -> Tuple[str, str]:
276
  device=0 if state.device == "cuda" else -1,
277
  truncation=True,
278
  padding="max_length",
279
- max_length=512
280
  )
281
  return f"βœ“ Tagger model loaded from {model_path}", ""
282
  except Exception as e:
@@ -336,8 +336,8 @@ def load_llm_model(model_path: str) -> Tuple[str, str]:
336
  state.llm_model = LLM(
337
  model=model_path,
338
  tensor_parallel_size=tp_size,
339
- gpu_memory_utilization=0.40,
340
- max_model_len=5000
341
  )
342
  state.llm_tokenizer = state.llm_model.get_tokenizer()
343
  return f"βœ“ LLM loaded from {model_path} (vLLM, tp={tp_size})", ""
 
276
  device=0 if state.device == "cuda" else -1,
277
  truncation=True,
278
  padding="max_length",
279
+ max_length=128
280
  )
281
  return f"βœ“ Tagger model loaded from {model_path}", ""
282
  except Exception as e:
 
336
  state.llm_model = LLM(
337
  model=model_path,
338
  tensor_parallel_size=tp_size,
339
+ gpu_memory_utilization=0.50,
340
+ max_model_len=15000
341
  )
342
  state.llm_tokenizer = state.llm_model.get_tokenizer()
343
  return f"βœ“ LLM loaded from {model_path} (vLLM, tp={tp_size})", ""