Merge ea517e3377 into 8a5029b121

opencompass/models/huggingface_above_v4_33.py

@@ -444,6 +444,7 @@ class HuggingFacewithChatTemplate(BaseModel):
             tokenize_kwargs['add_special_tokens'] = False
             tokens = self.tokenizer.batch_encode_plus(messages, **tokenize_kwargs)
 
+        prompt_list = messages
         tokens = {k: v.to(self.model.device) for k, v in tokens.items()}
 
         if self.mode == 'mid':
@@ -486,7 +487,7 @@ class HuggingFacewithChatTemplate(BaseModel):
         for stop in stopping_criteria:
             decodeds = [t.split(stop)[0] for t in decodeds]
 
-        return decodeds
+        return prompt_list, decodeds
 
     def get_token_len(self, prompt: str) -> int:
         m = _convert_chat_messages([prompt])[0]

opencompass/models/vllm_with_tf_above_v4_33.py

@@ -126,7 +126,7 @@ class VLLMwithChatTemplate(BaseModel):
             prompt_list.append(prompt)
             output_strs.append(generated_text)
 
-        return output_strs
+        return prompt_list, output_strs
 
     def get_token_len(self, prompt: str) -> int:
         """Get lengths of the tokenized strings.

opencompass/openicl/icl_inferencer/icl_base_inferencer.py

@@ -111,13 +111,20 @@ class GenInferencerOutputHandler:
         """Dump the result to a json file."""
         dump_results_dict(self.results_dict, Path(save_dir) / filename)
 
-    def save_results(self, origin_prompt, prediction, idx, gold=None):
+    def save_results(self,
-        self.results_dict[str(idx)] = {
+                     origin_prompt,
-            'origin_prompt': origin_prompt,
+                     prediction,
-            'prediction': prediction,
+                     idx,
-        }
+                     api_prompts=None,
+                     gold=None):
+        results = {}
+        if api_prompts:
+            results['api_prompts'] = api_prompts
+        results['origin_prompt'] = origin_prompt
+        results['prediction'] = prediction
         if gold:
-            self.results_dict[str(idx)]['gold'] = gold
+            results['gold'] = gold
+        self.results_dict[str(idx)] = results
 
 
 class PPLInferencerOutputHandler:

opencompass/openicl/icl_inferencer/icl_gen_inferencer.py

@@ -153,18 +153,25 @@ class GenInferencer(BaseInferencer):
                 results = self.model.generate_from_template(
                     entry, max_out_len=self.max_out_len, **extra_gen_kwargs)
                 generated = results
+                if isinstance(generated, tuple):
+                    api_prompts_list = parsed_entries
+                    prompts, generated = generated
+                else:
+                    api_prompts_list = [None] * len(generated)
+                    prompts = parsed_entries
 
             num_return_sequences = getattr(self.model, 'generation_kwargs',
                                            {}).get('num_return_sequences', 1)
             # 5-3. Save current output
-            for prompt, prediction, gold in zip(
+            for api_prompts, prompt, prediction, gold in zip(
-                    parsed_entries, batched(generated, num_return_sequences),
+                    api_prompts_list, prompts,
-                    golds):
+                    batched(generated, num_return_sequences), golds):
                 if num_return_sequences == 1:
                     prediction = prediction[0]
                 output_handler.save_results(prompt,
                                             prediction,
                                             index,
+                                            api_prompts=api_prompts,
                                             gold=gold)
                 index = index + 1