Switch to LLM2Vec 3.1 pair to fix Space 401 on gated Llama 3.0

kimodo/model/llm2vec/llm2vec.py

@@ -173,7 +173,10 @@ class LLM2Vec(nn.Module):
         return cls(model=model, tokenizer=tokenizer, **config)
 
     def prepare_for_tokenization(self, text):
-        if self.model.config._name_or_path == "meta-llama/Meta-Llama-3-8B-Instruct":
+        if self.model.config._name_or_path in [
+            "meta-llama/Meta-Llama-3-8B-Instruct",
+            "meta-llama/Meta-Llama-3.1-8B-Instruct",
+        ]:
             text = "<|start_header_id|>user<|end_header_id|>\n\n" + text.strip() + "<|eot_id|>"
             return text
         if self.model.config._name_or_path in [

kimodo/model/load_model.py

@@ -30,8 +30,8 @@ TEXT_ENCODER_PRESETS = {
     "llm2vec": {
         "target": "kimodo.model.LLM2VecEncoder",
         "kwargs": {
-            "base_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp",
-            "peft_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp-supervised",
+            "base_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-31-8B-Instruct-mntp",
+            "peft_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-31-8B-Instruct-mntp-supervised",
             "dtype": "bfloat16",
             "llm_dim": 4096,
         },

kimodo/scripts/run_text_encoder_server.py

@@ -22,8 +22,8 @@ TEXT_ENCODER_PRESETS = {
     "llm2vec": {
         "target": "kimodo.model.LLM2VecEncoder",
         "kwargs": {
-            "base_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp",
-            "peft_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp-supervised",
+            "base_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-31-8B-Instruct-mntp",
+            "peft_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-31-8B-Instruct-mntp-supervised",
             "dtype": "bfloat16",
             "llm_dim": 4096,
         },

kimodo/scripts/text_encoder_health.py

@@ -12,8 +12,8 @@ from transformers import AutoConfig
 
 TEXT_ENCODER_PRESETS = {
     "llm2vec": {
-        "base_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp",
-        "peft_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp-supervised",
+        "base_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-31-8B-Instruct-mntp",
+        "peft_model_name_or_path": "McGill-NLP/LLM2Vec-Meta-Llama-31-8B-Instruct-mntp-supervised",
     }
 }