add monkey-chat

vlmeval/config.py

@@ -46,6 +46,7 @@
     'emu2':partial(Emu, name='emu2'),
     'emu2_chat':partial(Emu, name='emu2_chat'),
     'monkey':partial(Monkey, model_path='echo840/Monkey'),
+    'monkey-chat':partial(MonkeyChat, model_path='echo840/Monkey-Chat'),
 }
 
 api_models = {

vlmeval/vlm/__init__.py

@@ -16,4 +16,4 @@
 from .cogvlm import CogVlm
 from .sharedcaptioner import SharedCaptioner
 from .emu import Emu
-from .monkey import Monkey
+from .monkey import Monkey, MonkeyChat

vlmeval/vlm/monkey.py

@@ -22,7 +22,7 @@ def __init__(self, model_path='echo840/Monkey', **kwargs):
         torch.cuda.empty_cache()
 
     def generate(self, image_path, prompt, dataset=None):
-        cur_prompt = f'<img>{image_path}</img>\n{prompt} Answer:'
+        cur_prompt = f'<img>{image_path}</img> {prompt} Answer:'
         input_ids = self.tokenizer(cur_prompt, return_tensors='pt', padding='longest')
         attention_mask = input_ids.attention_mask
         input_ids = input_ids.input_ids
@@ -44,3 +44,47 @@ def generate(self, image_path, prompt, dataset=None):
         response = self.tokenizer.decode(output_ids[0][input_ids.size(1):].cpu(), skip_special_tokens=True).strip()
 
         return response
+
+class MonkeyChat:
+
+    INSTALL_REQ = False
+
+    def __init__(self, model_path='echo840/Monkey-Chat', **kwargs):
+        assert model_path is not None
+        self.model_path = model_path
+        self.tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
+        self.model = AutoModelForCausalLM.from_pretrained(model_path, device_map='cuda', trust_remote_code=True).eval()
+        self.kwargs = kwargs
+
+        self.tokenizer.padding_side = 'left'
+        self.tokenizer.pad_token_id = self.tokenizer.eod_id
+
+        warnings.warn(f"Following kwargs received: {self.kwargs}, will use as generation config. ")
+        torch.cuda.empty_cache()
+
+    def generate(self, image_path, prompt, dataset=None):
+        if dataset == 'MMVet':
+            cur_prompt = f'<img>{image_path}</img> {prompt} Answer: '
+        else:
+            cur_prompt = f'<img>{image_path}</img> \n {prompt} Answer: '
+        input_ids = self.tokenizer(cur_prompt, return_tensors='pt', padding='longest')
+        attention_mask = input_ids.attention_mask
+        input_ids = input_ids.input_ids
+
+        output_ids = self.model.generate(
+                input_ids=input_ids.cuda(),
+                attention_mask=attention_mask.cuda(),
+                do_sample=False,
+                num_beams=1,
+                max_new_tokens=10,
+                min_new_tokens=1,
+                length_penalty=1,
+                num_return_sequences=1,
+                output_hidden_states=True,
+                use_cache=True,
+                pad_token_id=self.tokenizer.eod_id,
+                eos_token_id=self.tokenizer.eod_id,
+            )
+        response = self.tokenizer.decode(output_ids[0][input_ids.size(1):].cpu(), skip_special_tokens=True).strip()
+
+        return response