Fix llm_int8_skip_modules

This commit is contained in:
Daniel Han-Chen 2024-02-06 18:53:32 +11:00
commit d0b1144423

View file

@ -128,7 +128,7 @@ class FastLanguageModel(FastLlamaModel):
"bnb_4bit_use_double_quant" : True,
"llm_int8_enable_fp32_cpu_offload" : False,
"llm_int8_has_fp16_weight" : False,
"llm_int8_skip_modules" : "null",
"llm_int8_skip_modules" : None,
"llm_int8_threshold" : 6.0,
"load_in_4bit" : True,
"load_in_8bit" : False,