support gemma 3

Open devops724 opened this issue 9 months ago • 2 comments

on run quantize in gemma3 i get this error Traceback (most recent call last): File "/home/user/workspace/api/document/quantize.py", line 9, in model = AutoAWQForCausalLM.from_pretrained(model_path) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/miniconda3/lib/python3.12/site-packages/awq/models/auto.py", line 76, in from_pretrained model_type = check_and_get_model_type( ^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/miniconda3/lib/python3.12/site-packages/awq/models/auto.py", line 51, in check_and_get_model_type raise TypeError(f"{config.model_type} isn't supported yet.") TypeError: gemma3 isn't supported yet.

Mar 22 '25 23:03 devops724

Apr 08 '25 10:04 ignaceHelsen

Apr 25 '25 04:04 Sumanth197