Switch LLM2Vec to Meta-Llama-3.1-8B-Instruct (accessible with current token)

app.py

@@ -28,7 +28,7 @@ os.environ.setdefault("HF_MODE", "1")
 # Avoid local LLM2Vec fallback on Spaces (requires gated Llama weights).
 os.environ.setdefault("TEXT_ENCODER_MODE", "api")
 os.environ.setdefault("TEXT_ENCODER", "llm2vec")
-os.environ.setdefault("LLM2VEC_BASE_MODEL", "rydlrKE/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp")
+os.environ.setdefault("LLM2VEC_BASE_MODEL", "meta-llama/Meta-Llama-3.1-8B-Instruct")
 os.environ.setdefault(
     "LLM2VEC_PEFT_MODEL",
     "McGill-NLP/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp-supervised",

kimodo/model/load_model.py

@@ -20,7 +20,7 @@ from .loading import (
 from .registry import get_model_info, resolve_model_name
 
 DEFAULT_TEXT_ENCODER = "llm2vec"
-DEFAULT_LLM2VEC_BASE = "rydlrKE/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp"
+DEFAULT_LLM2VEC_BASE = "meta-llama/Meta-Llama-3.1-8B-Instruct"
 DEFAULT_LLM2VEC_PEFT = "McGill-NLP/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp-supervised"
 TEXT_ENCODER_PRESETS = {
     "llm2vec": {

kimodo/scripts/run_text_encoder_server.py

@@ -17,7 +17,7 @@ DEFAULT_SERVER_NAME = "0.0.0.0"
 DEFAULT_SERVER_PORT = 9550
 DEFAULT_TMP_FOLDER = "/tmp/text_encoder/"
 DEFAULT_TEXT_ENCODER = "llm2vec"
-DEFAULT_LLM2VEC_BASE = "rydlrKE/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp"
+DEFAULT_LLM2VEC_BASE = "meta-llama/Meta-Llama-3.1-8B-Instruct"
 DEFAULT_LLM2VEC_PEFT = "McGill-NLP/LLM2Vec-Meta-Llama-3-8B-Instruct-mntp-supervised"
 TEXT_ENCODER_PRESETS = {
     "llm2vec": {