Upload folder using huggingface_hub

modeling_internvl_chat.py

@@ -35,6 +35,7 @@ def version_cmp(v1, v2, op='eq'):
 class InternVLChatModel(PreTrainedModel):
     config_class = InternVLChatConfig
     main_input_name = 'pixel_values'
+    base_model_prefix = 'language_model'
     _supports_flash_attn_2 = True
     _no_split_modules = ['InternVisionModel', 'LlamaDecoderLayer', 'Phi3DecoderLayer']
 
@@ -101,7 +102,7 @@ class InternVLChatModel(PreTrainedModel):
         return_dict = return_dict if return_dict is not None else self.config.use_return_dict
 
         image_flags = image_flags.squeeze(-1)
-        input_embeds = self.language_model.get_input_embeddings()(input_ids)
+        input_embeds = self.language_model.get_input_embeddings()(input_ids).clone()
 
         vit_embeds = self.extract_feature(pixel_values)
         vit_embeds = vit_embeds[image_flags == 1]
@@ -234,8 +235,8 @@ class InternVLChatModel(PreTrainedModel):
 
         tokenizer.padding_side = 'left'
         model_inputs = tokenizer(queries, return_tensors='pt', padding=True)
-        input_ids = model_inputs['input_ids'].cuda()
-        attention_mask = model_inputs['attention_mask'].cuda()
+        input_ids = model_inputs['input_ids'].to(self.device)
+        attention_mask = model_inputs['attention_mask'].to(self.device)
         eos_token_id = tokenizer.convert_tokens_to_ids(template.sep)
         generation_config['eos_token_id'] = eos_token_id
         generation_output = self.generate(
@@ -283,8 +284,8 @@ class InternVLChatModel(PreTrainedModel):
             query = query.replace('<image>', image_tokens, 1)
 
         model_inputs = tokenizer(query, return_tensors='pt')
-        input_ids = model_inputs['input_ids'].cuda()
-        attention_mask = model_inputs['attention_mask'].cuda()
+        input_ids = model_inputs['input_ids'].to(self.device)
+        attention_mask = model_inputs['attention_mask'].to(self.device)
         generation_config['eos_token_id'] = eos_token_id
         generation_output = self.generate(
             pixel_values=pixel_values,

modeling_phi3.py

@@ -1370,7 +1370,7 @@ class Phi3ForCausalLM(Phi3PreTrainedModel):
                 position_ids = position_ids[:, -input_ids.shape[1] :]
 
         # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
-        if inputs_embeds is not None and past_key_values is None:
+        if (inputs_embeds is not None and past_key_values is None) or (inputs_embeds is not None and len(past_key_values) == 0):
             model_inputs = {'inputs_embeds': inputs_embeds}
         else:
             model_inputs = {'input_ids': input_ids}