Cleanup the text part of the clip model as redundant as it's not even used

configs/clip/special_tokens_map.json

@@ -1,15 +0,0 @@
-{
-  "bos_token": "<s>",
-  "cls_token": "<s>",
-  "eos_token": "</s>",
-  "mask_token": {
-    "content": "<mask>",
-    "lstrip": true,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  },
-  "pad_token": "<pad>",
-  "sep_token": "</s>",
-  "unk_token": "<unk>"
-}

configs/clip/tokenizer_config.json

@@ -1,19 +0,0 @@
-{
-  "bos_token": "<s>",
-  "clean_up_tokenization_spaces": true,
-  "cls_token": "<s>",
-  "eos_token": "</s>",
-  "mask_token": {
-    "__type": "AddedToken",
-    "content": "<mask>",
-    "lstrip": true,
-    "normalized": true,
-    "rstrip": false,
-    "single_word": false
-  },
-  "model_max_length": 512,
-  "pad_token": "<pad>",
-  "sep_token": "</s>",
-  "tokenizer_class": "XLMRobertaTokenizer",
-  "unk_token": "<unk>"
-}

configs/clip/xlm-roberta-large%2Fspecial_tokens_map.json

@@ -1,15 +0,0 @@
-{
-  "bos_token": "<s>",
-  "cls_token": "<s>",
-  "eos_token": "</s>",
-  "mask_token": {
-    "content": "<mask>",
-    "lstrip": true,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  },
-  "pad_token": "<pad>",
-  "sep_token": "</s>",
-  "unk_token": "<unk>"
-}

nodes.py

@@ -641,13 +641,11 @@ class LoadWanVideoClipTextEncoder:
 
         text_encoder_load_device = device if load_device == "main_device" else offload_device
 
-        tokenizer_path = os.path.join(script_directory, "configs", "clip")
-
         dtype = {"bf16": torch.bfloat16, "fp16": torch.float16, "fp32": torch.float32}[precision]
 
         model_path = folder_paths.get_full_path("text_encoders", model_name)
         sd = load_torch_file(model_path, safe_load=True)
-        clip_model = CLIPModel(dtype=dtype, device=device, state_dict=sd, tokenizer_path=tokenizer_path)
+        clip_model = CLIPModel(dtype=dtype, device=device, state_dict=sd)
         clip_model.model.to(text_encoder_load_device)
         del sd
         

wanvideo/modules/clip.py

@@ -393,17 +393,6 @@ class XLMRobertaCLIP(nn.Module):
             proj_dropout=proj_dropout,
             embedding_dropout=embedding_dropout,
             norm_eps=norm_eps)
-        # self.textual = XLMRobertaWithHead(
-        #     vocab_size=vocab_size,
-        #     max_seq_len=max_text_len,
-        #     type_size=type_size,
-        #     pad_id=pad_id,
-        #     dim=text_dim,
-        #     out_dim=embed_dim,
-        #     num_heads=text_heads,
-        #     num_layers=text_layers,
-        #     post_norm=text_post_norm,
-        #     dropout=text_dropout)
         self.log_scale = nn.Parameter(math.log(1 / 0.07) * torch.ones([]))
 
     def forward(self, imgs, txt_ids):
@@ -503,10 +492,9 @@ def clip_xlm_roberta_vit_h_14(
 
 class CLIPModel:
 
-    def __init__(self, dtype, device, state_dict, tokenizer_path):
+    def __init__(self, dtype, device, state_dict):
         self.dtype = dtype
         self.device = device
-        self.tokenizer_path = tokenizer_path
 
         # init model
         with init_empty_weights():
@@ -521,13 +509,6 @@ class CLIPModel:
 
         for name, param in self.model.named_parameters():
             set_module_tensor_to_device(self.model, name, device=device, dtype=dtype, value=state_dict[name])
-        #self.model.load_state_dict(state_dict)
-
-        # init tokenizer
-        self.tokenizer = HuggingfaceTokenizer(
-            name=tokenizer_path,
-            seq_len=self.model.max_text_len - 2,
-            clean='whitespace')
 
     def visual(self, image):
         # forward