Error when trying to fine-tune the new Llama-3.1

#9
by GuusBouwensNL - opened

I just build the autotrain advanced and loaded my training dataset and the llama-3.1-70B-Instruct model and was starting the DPOfine tuning, but then this error occurred:

"train has failed due to an exception: Traceback (most recent call last): File "/app/env/lib/python3.10/site-packages/autotrain/trainers/common.py", line 117, in wrapper return func(*args, kwargs) File "/app/env/lib/python3.10/site-packages/autotrain/trainers/clm/main.py"
....
LLAMA_ATTENTION_CLASSES[config._attn_implementation](config=config, layer_idx=layer_idx) File "/app/env/lib/python3.10/site-packages/transformers/models/llama/modeling_llama.py", line 306, in init self.rotary_emb = LlamaRotaryEmbedding(config=self.config) File "/app/env/lib/python3.10/site-packages/transformers/models/llama/modeling_llama.py", line 110, in init self.rope_type = config.rope_scaling.get("rope_type", config.rope_scaling["type"]) KeyError: 'type' ERROR | 2024-07-23 20:45:55 | autotrain.trainers.common:wrapper:121 - 'type'"

I guess this means the config file for 3.1 has not been fully updated?

nvm just seen last post

GuusBouwensNL changed discussion status to closed

Sign up or log in to comment