llama loaders no longer inherit ModelHandler

Author: Jonathan

llm_eval/llama.py

@@ -3,7 +3,7 @@
 from transformers import AutoTokenizer, AutoModelForCausalLM
 
 from .handler import ModelHandler
-class Llama3ModelHandler(ModelHandler):
+class Llama3ModelHandler():
       def load_model_and_tokenizer(self, device, model_id):
         tokenizer = AutoTokenizer.from_pretrained(model_id)
         if tokenizer.pad_token is None:
@@ -18,7 +18,7 @@ def load_model_and_tokenizer(self, device, model_id):
         return tokenizer, model
 
 from .handler import ModelHandler
-class Llama2ModelHandler(ModelHandler):
+class Llama2ModelHandler():
       def load_model_and_tokenizer(self, device, model_id):
         tokenizer = AutoTokenizer.from_pretrained(model_id)
         if tokenizer.pad_token is None:

pyproject.toml

@@ -6,7 +6,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "llm-eval"
-version = "0.6.7"
+version = "0.6.8"
 authors = [
     {name = "Jonathan Eisenzopf", email = "[email protected]"},
 ]