From e08e0e5814bf4245d3a3ed48d9a0d93d230236db Mon Sep 17 00:00:00 2001
From: "yuze.zyz" <yuze.zyz@alibaba-inc.com>
Date: Wed, 29 Nov 2023 20:36:55 +0800
Subject: [PATCH 1/6] support ms

Former-commit-id: d38a2e7341100902b6c761895b1fe6191c905d06
---
 src/llmtuner/extras/constants.py   | 172 +++++++++++++++++++++++++++++
 src/llmtuner/hparams/model_args.py |   3 +-
 src/llmtuner/model/loader.py       |  14 +++
 3 files changed, 188 insertions(+), 1 deletion(-)

diff --git a/src/llmtuner/extras/constants.py b/src/llmtuner/extras/constants.py
index 3257678e..307f3c97 100644
--- a/src/llmtuner/extras/constants.py
+++ b/src/llmtuner/extras/constants.py
@@ -1,3 +1,4 @@
+import os
 from collections import defaultdict, OrderedDict
 from typing import Dict, Optional
 
@@ -20,6 +21,8 @@ SUBJECTS = ["Average", "STEM", "Social Sciences", "Humanities", "Other"]
 
 SUPPORTED_MODELS = OrderedDict()
 
+MODELSCOPE_MODELS = OrderedDict()
+
 TRAINING_STAGES = {
     "Supervised Fine-Tuning": "sft",
     "Reward Modeling": "rm",
@@ -40,7 +43,11 @@ def register_model_group(
             prefix = name.split("-")[0]
         else:
             assert prefix == name.split("-")[0], "prefix should be identical."
+
         SUPPORTED_MODELS[name] = path
+        if os.environ.get('USE_MODELSCOPE_HUB', False) and name in MODELSCOPE_MODELS:
+            # Use ModelScope modelhub
+            SUPPORTED_MODELS[name] = MODELSCOPE_MODELS[name]
     if module is not None:
         DEFAULT_MODULE[prefix] = module
     if template is not None:
@@ -58,6 +65,13 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "Baichuan-7B-Base": "baichuan-inc/baichuan-7B",
+    "Baichuan-13B-Base": "baichuan-inc/Baichuan-13B-Base",
+    "Baichuan-13B-Chat": "baichuan-inc/Baichuan-13B-Base"
+})
+
+
 register_model_group(
     models={
         "Baichuan2-7B-Base": "baichuan-inc/Baichuan2-7B-Base",
@@ -70,6 +84,14 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "Baichuan2-7B-Base": "baichuan-inc/Baichuan2-7B-Base",
+    "Baichuan2-13B-Base": "baichuan-inc/Baichuan2-13B-Base",
+    "Baichuan2-7B-Chat": "baichuan-inc/Baichuan2-7B-Chat",
+    "Baichuan2-13B-Chat": "baichuan-inc/Baichuan2-13B-Chat"
+})
+
+
 register_model_group(
     models={
         "BLOOM-560M": "bigscience/bloom-560m",
@@ -80,6 +102,13 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "BLOOM-560M": "AI-ModelScope/bloom-560m",
+    "BLOOM-3B": "bigscience/bloom-3b",
+    "BLOOM-7B1": "bigscience/bloom-7b1"
+})
+
+
 register_model_group(
     models={
         "BLOOMZ-560M": "bigscience/bloomz-560m",
@@ -90,6 +119,13 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "BLOOMZ-560M": "bigscience/bloomz-560m",
+    "BLOOMZ-3B": "bigscience/bloomz-3b",
+    "BLOOMZ-7B1-mt": "AI-ModelScope/bloomz-7b1-mt"
+})
+
+
 register_model_group(
     models={
         "BlueLM-7B-Base": "vivo-ai/BlueLM-7B-Base",
@@ -99,6 +135,12 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "BlueLM-7B-Base": "vivo-ai/BlueLM-7B-Base",
+    "BlueLM-7B-Chat": "vivo-ai/BlueLM-7B-Chat"
+})
+
+
 register_model_group(
     models={
         "ChatGLM2-6B-Chat": "THUDM/chatglm2-6b"
@@ -108,6 +150,11 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "ChatGLM2-6B-Chat": "ZhipuAI/chatglm2-6b"
+})
+
+
 register_model_group(
     models={
         "ChatGLM3-6B-Base": "THUDM/chatglm3-6b-base",
@@ -118,6 +165,12 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "ChatGLM3-6B-Base": "ZhipuAI/chatglm3-6b-base",
+    "ChatGLM3-6B-Chat": "ZhipuAI/chatglm3-6b"
+})
+
+
 register_model_group(
     models={
         "ChineseLLaMA2-1.3B": "hfl/chinese-llama-2-1.3b",
@@ -131,6 +184,16 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "ChineseLLaMA2-1.3B": "hfl/chinese-llama-2-1.3b",
+    "ChineseLLaMA2-7B": "hfl/chinese-llama-2-7b",
+    "ChineseLLaMA2-13B": "hfl/chinese-llama-2-13b",
+    "ChineseLLaMA2-1.3B-Chat": "hfl/chinese-alpaca-2-1.3b",
+    "ChineseLLaMA2-7B-Chat": "hfl/chinese-alpaca-2-7b",
+    "ChineseLLaMA2-13B-Chat": "hfl/chinese-alpaca-2-13b"
+})
+
+
 register_model_group(
     models={
         "Falcon-7B": "tiiuae/falcon-7b",
@@ -145,6 +208,16 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "Falcon-7B": "tiiuae/falcon-7b",
+    "Falcon-40B": "tiiuae/falcon-40b",
+    "Falcon-180B": "tiiuae/falcon-180B",
+    "Falcon-7B-Chat": "AI-ModelScope/falcon-7b-instruct",
+    "Falcon-40B-Chat": "tiiuae/falcon-40b-instruct",
+    "Falcon-180B-Chat": "tiiuae/falcon-180B-chat"
+})
+
+
 register_model_group(
     models={
         "InternLM-7B": "internlm/internlm-7b",
@@ -156,6 +229,14 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "InternLM-7B": "Shanghai_AI_Laboratory/internlm-7b",
+    "InternLM-20B": "Shanghai_AI_Laboratory/internlm-20b",
+    "InternLM-7B-Chat": "Shanghai_AI_Laboratory/internlm-chat-7b",
+    "InternLM-20B-Chat": "Shanghai_AI_Laboratory/internlm-chat-20b"
+})
+
+
 register_model_group(
     models={
         "LingoWhale-8B": "deeplang-ai/LingoWhale-8B"
@@ -164,6 +245,11 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "LingoWhale-8B": "DeepLang/LingoWhale-8B"
+})
+
+
 register_model_group(
     models={
         "LLaMA-7B": "huggyllama/llama-7b",
@@ -174,6 +260,14 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "LLaMA-7B": "skyline2006/llama-7b",
+    "LLaMA-13B": "skyline2006/llama-13b",
+    "LLaMA-30B": "skyline2006/llama-30b",
+    "LLaMA-65B": "skyline2006/llama-65b"
+})
+
+
 register_model_group(
     models={
         "LLaMA2-7B": "meta-llama/Llama-2-7b-hf",
@@ -187,6 +281,16 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "LLaMA2-7B": "modelscope/Llama-2-7b-ms",
+    "LLaMA2-13B": "modelscope/Llama-2-13b-ms",
+    "LLaMA2-70B": "modelscope/Llama-2-70b-ms",
+    "LLaMA2-7B-Chat": "modelscope/Llama-2-7b-chat-ms",
+    "LLaMA2-13B-Chat": "modelscope/Llama-2-13b-chat-ms",
+    "LLaMA2-70B-Chat": "modelscope/Llama-2-70b-chat-ms"
+})
+
+
 register_model_group(
     models={
         "Mistral-7B": "mistralai/Mistral-7B-v0.1",
@@ -196,6 +300,12 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "Mistral-7B": "AI-ModelScope/Mistral-7B-v0.1",
+    "Mistral-7B-Chat": "AI-ModelScope/Mistral-7B-Instruct-v0.1"
+})
+
+
 register_model_group(
     models={
         "OpenChat3.5-7B-Chat": "openchat/openchat_3.5"
@@ -204,6 +314,11 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "OpenChat3.5-7B-Chat": "openchat/openchat_3.5"
+})
+
+
 register_model_group(
     models={
         "Phi1.5-1.3B": "microsoft/phi-1_5"
@@ -212,6 +327,11 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "Phi1.5-1.3B": "microsoft/phi-1_5"
+})
+
+
 register_model_group(
     models={
         "Qwen-7B": "Qwen/Qwen-7B",
@@ -228,6 +348,18 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "Qwen-7B": "qwen/Qwen-7B",
+    "Qwen-14B": "qwen/Qwen-14B",
+    "Qwen-7B-Chat": "qwen/Qwen-7B-Chat",
+    "Qwen-14B-Chat": "qwen/Qwen-14B-Chat",
+    "Qwen-7B-int8-Chat": "qwen/Qwen-7B-Chat-Int8",
+    "Qwen-7B-int4-Chat": "qwen/Qwen-7B-Chat-Int4",
+    "Qwen-14B-int8-Chat": "qwen/Qwen-14B-Chat-Int8",
+    "Qwen-14B-int4-Chat": "qwen/Qwen-14B-Chat-Int4"
+})
+
+
 register_model_group(
     models={
         "Skywork-13B-Base": "Skywork/Skywork-13B-base"
@@ -235,6 +367,11 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "Skywork-13B-Base": "skywork/Skywork-13B-base"
+})
+
+
 register_model_group(
     models={
         "Vicuna1.5-7B-Chat": "lmsys/vicuna-7b-v1.5",
@@ -244,6 +381,12 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "Vicuna1.5-7B-Chat": "AI-ModelScope/vicuna-7b-v1.5",
+    "Vicuna1.5-13B-Chat": "lmsys/vicuna-13b-v1.5"
+})
+
+
 register_model_group(
     models={
         "XVERSE-7B": "xverse/XVERSE-7B",
@@ -256,6 +399,15 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "XVERSE-7B": "xverse/XVERSE-7B",
+    "XVERSE-13B": "xverse/XVERSE-13B",
+    "XVERSE-65B": "xverse/XVERSE-65B",
+    "XVERSE-7B-Chat": "xverse/XVERSE-7B-Chat",
+    "XVERSE-13B-Chat": "xverse/XVERSE-13B-Chat"
+})
+
+
 register_model_group(
     models={
         "Yayi-7B": "wenge-research/yayi-7b-llama2",
@@ -265,6 +417,12 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "Yayi-7B": "wenge-research/yayi-7b-llama2",
+    "Yayi-13B": "wenge-research/yayi-13b-llama2"
+})
+
+
 register_model_group(
     models={
         "Yi-6B": "01-ai/Yi-6B",
@@ -276,6 +434,14 @@ register_model_group(
 )
 
 
+MODELSCOPE_MODELS.update({
+    "Yi-6B": "01ai/Yi-6B",
+    "Yi-34B": "01ai/Yi-34B",
+    "Yi-34B-Chat": "01ai/Yi-34B-Chat",
+    "Yi-34B-int8-Chat": "01ai/Yi-34B-Chat-8bits"
+})
+
+
 register_model_group(
     models={
         "Zephyr-7B-Alpha-Chat": "HuggingFaceH4/zephyr-7b-alpha",
@@ -283,3 +449,9 @@ register_model_group(
     },
     template="zephyr"
 )
+
+
+MODELSCOPE_MODELS.update({
+    "Zephyr-7B-Alpha-Chat": "HuggingFaceH4/zephyr-7b-alpha",
+    "Zephyr-7B-Beta-Chat": "modelscope/zephyr-7b-beta"
+})
diff --git a/src/llmtuner/hparams/model_args.py b/src/llmtuner/hparams/model_args.py
index 4bda39d5..ebf6cafa 100644
--- a/src/llmtuner/hparams/model_args.py
+++ b/src/llmtuner/hparams/model_args.py
@@ -8,7 +8,8 @@ class ModelArguments:
     Arguments pertaining to which model/config/tokenizer we are going to fine-tune.
     """
     model_name_or_path: str = field(
-        metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models."}
+        metadata={"help": "Path to pretrained model or model identifier "
+                          "from huggingface.co/models or modelscope.cn/models."}
     )
     cache_dir: Optional[str] = field(
         default=None,
diff --git a/src/llmtuner/model/loader.py b/src/llmtuner/model/loader.py
index 5561837a..5fed514a 100644
--- a/src/llmtuner/model/loader.py
+++ b/src/llmtuner/model/loader.py
@@ -1,4 +1,6 @@
 import math
+import os
+
 import torch
 from types import MethodType
 from typing import TYPE_CHECKING, Literal, Optional, Tuple
@@ -63,6 +65,8 @@ def load_model_and_tokenizer(
         "token": model_args.hf_hub_token
     }
 
+    try_download_model_from_ms(model_args)
+
     tokenizer = AutoTokenizer.from_pretrained(
         model_args.model_name_or_path,
         use_fast=model_args.use_fast_tokenizer,
@@ -228,3 +232,13 @@ def load_model_and_tokenizer(
         logger.info("This IS expected that the trainable params is 0 if you are using model for inference only.")
 
     return model, tokenizer
+
+
+def try_download_model_from_ms(model_args):
+    if os.environ.get('USE_MODELSCOPE_HUB', False) and not os.path.exists(model_args.model_name_or_path):
+        try:
+            from modelscope import snapshot_download
+            model_args.model_name_or_path = snapshot_download(model_args.model_name_or_path, model_args.model_revision)
+        except ImportError as e:
+            raise ImportError(f'You are using `USE_MODELSCOPE_HUB=True` but you have no modelscope sdk installed. '
+                              f'Please install it by `pip install modelscope -U`') from e

From 45925e4a9ce43d5f100175cd6a38a32bff626f08 Mon Sep 17 00:00:00 2001
From: "yuze.zyz" <yuze.zyz@alibaba-inc.com>
Date: Wed, 29 Nov 2023 21:43:58 +0800
Subject: [PATCH 2/6] fix

Former-commit-id: fb2204c183ae8c061ed6ec7f4f1bfbb0b4900c9b
---
 src/llmtuner/extras/constants.py | 543 +++++++++++++++++--------------
 src/llmtuner/model/loader.py     |   5 +-
 2 files changed, 303 insertions(+), 245 deletions(-)

diff --git a/src/llmtuner/extras/constants.py b/src/llmtuner/extras/constants.py
index 307f3c97..9de07bc2 100644
--- a/src/llmtuner/extras/constants.py
+++ b/src/llmtuner/extras/constants.py
@@ -1,7 +1,6 @@
 import os
 from collections import defaultdict, OrderedDict
-from typing import Dict, Optional
-
+from typing import Dict, Optional, Union
 
 CHOICES = ["A", "B", "C", "D"]
 
@@ -33,7 +32,7 @@ TRAINING_STAGES = {
 
 
 def register_model_group(
-    models: Dict[str, str],
+    models: Dict[str, Union[str, Dict[str, str]]],
     module: Optional[str] = None,
     template: Optional[str] = None
 ) -> None:
@@ -44,10 +43,12 @@ def register_model_group(
         else:
             assert prefix == name.split("-")[0], "prefix should be identical."
 
-        SUPPORTED_MODELS[name] = path
-        if os.environ.get('USE_MODELSCOPE_HUB', False) and name in MODELSCOPE_MODELS:
+        if not os.environ.get('USE_MODELSCOPE_HUB', False):
+            # If path is a string, we treat it as a huggingface model-id by default.
+            SUPPORTED_MODELS[name] = path["hf"] if isinstance(path, dict) else path
+        elif isinstance(path, dict) and "ms" in path:
             # Use ModelScope modelhub
-            SUPPORTED_MODELS[name] = MODELSCOPE_MODELS[name]
+            SUPPORTED_MODELS[name] = path["ms"]
     if module is not None:
         DEFAULT_MODULE[prefix] = module
     if template is not None:
@@ -56,402 +57,456 @@ def register_model_group(
 
 register_model_group(
     models={
-        "Baichuan-7B-Base": "baichuan-inc/Baichuan-7B",
-        "Baichuan-13B-Base": "baichuan-inc/Baichuan-13B-Base",
-        "Baichuan-13B-Chat": "baichuan-inc/Baichuan-13B-Chat"
+        "Baichuan-7B-Base": {
+            "hf": "baichuan-inc/Baichuan-7B",
+            "ms": "baichuan-inc/baichuan-7B",
+        },
+        "Baichuan-13B-Base": {
+            "hf": "baichuan-inc/Baichuan-13B-Base",
+            "ms": "baichuan-inc/Baichuan-13B-Base",
+        },
+        "Baichuan-13B-Chat": {
+            "hf": "baichuan-inc/Baichuan-13B-Chat",
+            "ms": "baichuan-inc/Baichuan-13B-Base",
+        }
     },
     module="W_pack",
     template="baichuan"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "Baichuan-7B-Base": "baichuan-inc/baichuan-7B",
-    "Baichuan-13B-Base": "baichuan-inc/Baichuan-13B-Base",
-    "Baichuan-13B-Chat": "baichuan-inc/Baichuan-13B-Base"
-})
-
-
 register_model_group(
     models={
-        "Baichuan2-7B-Base": "baichuan-inc/Baichuan2-7B-Base",
-        "Baichuan2-13B-Base": "baichuan-inc/Baichuan2-13B-Base",
-        "Baichuan2-7B-Chat": "baichuan-inc/Baichuan2-7B-Chat",
-        "Baichuan2-13B-Chat": "baichuan-inc/Baichuan2-13B-Chat"
+        "Baichuan2-7B-Base": {
+            "hf": "baichuan-inc/Baichuan2-7B-Base",
+            "ms": "baichuan-inc/Baichuan2-7B-Base",
+        },
+        "Baichuan2-13B-Base": {
+            "hf": "baichuan-inc/Baichuan2-13B-Base",
+            "ms": "baichuan-inc/Baichuan2-13B-Base",
+        },
+        "Baichuan2-7B-Chat": {
+            "hf": "baichuan-inc/Baichuan2-7B-Chat",
+            "ms": "baichuan-inc/Baichuan2-7B-Chat",
+        },
+        "Baichuan2-13B-Chat": {
+            "hf": "baichuan-inc/Baichuan2-13B-Chat",
+            "ms": "baichuan-inc/Baichuan2-13B-Chat",
+        }
     },
     module="W_pack",
     template="baichuan2"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "Baichuan2-7B-Base": "baichuan-inc/Baichuan2-7B-Base",
-    "Baichuan2-13B-Base": "baichuan-inc/Baichuan2-13B-Base",
-    "Baichuan2-7B-Chat": "baichuan-inc/Baichuan2-7B-Chat",
-    "Baichuan2-13B-Chat": "baichuan-inc/Baichuan2-13B-Chat"
-})
-
-
 register_model_group(
     models={
-        "BLOOM-560M": "bigscience/bloom-560m",
-        "BLOOM-3B": "bigscience/bloom-3b",
-        "BLOOM-7B1": "bigscience/bloom-7b1"
+        "BLOOM-560M": {
+            "hf": "bigscience/bloom-560m",
+            "ms": "AI-ModelScope/bloom-560m",
+        },
+        "BLOOM-3B": {
+            "hf": "bigscience/bloom-3b",
+            "ms": "bigscience/bloom-3b",
+        },
+        "BLOOM-7B1": {
+            "hf": "bigscience/bloom-7b1",
+            "ms": "bigscience/bloom-7b1",
+        }
     },
     module="query_key_value"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "BLOOM-560M": "AI-ModelScope/bloom-560m",
-    "BLOOM-3B": "bigscience/bloom-3b",
-    "BLOOM-7B1": "bigscience/bloom-7b1"
-})
-
-
 register_model_group(
     models={
-        "BLOOMZ-560M": "bigscience/bloomz-560m",
-        "BLOOMZ-3B": "bigscience/bloomz-3b",
-        "BLOOMZ-7B1-mt": "bigscience/bloomz-7b1-mt"
+        "BLOOMZ-560M": {
+            "hf": "bigscience/bloomz-560m",
+            "ms": "bigscience/bloomz-560m",
+        },
+        "BLOOMZ-3B": {
+            "hf": "bigscience/bloomz-3b",
+            "ms": "bigscience/bloomz-3b",
+        },
+        "BLOOMZ-7B1-mt": {
+            "hf": "bigscience/bloomz-7b1-mt",
+            "ms": "AI-ModelScope/bloomz-7b1-mt",
+        }
     },
     module="query_key_value"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "BLOOMZ-560M": "bigscience/bloomz-560m",
-    "BLOOMZ-3B": "bigscience/bloomz-3b",
-    "BLOOMZ-7B1-mt": "AI-ModelScope/bloomz-7b1-mt"
-})
-
-
 register_model_group(
     models={
-        "BlueLM-7B-Base": "vivo-ai/BlueLM-7B-Base",
-        "BlueLM-7B-Chat": "vivo-ai/BlueLM-7B-Chat"
+        "BlueLM-7B-Base": {
+            "hf": "vivo-ai/BlueLM-7B-Base",
+            "ms": "vivo-ai/BlueLM-7B-Base",
+        },
+        "BlueLM-7B-Chat": {
+            "hf": "vivo-ai/BlueLM-7B-Chat",
+            "ms": "vivo-ai/BlueLM-7B-Chat",
+        }
     },
     template="bluelm"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "BlueLM-7B-Base": "vivo-ai/BlueLM-7B-Base",
-    "BlueLM-7B-Chat": "vivo-ai/BlueLM-7B-Chat"
-})
-
-
 register_model_group(
     models={
-        "ChatGLM2-6B-Chat": "THUDM/chatglm2-6b"
+        "ChatGLM2-6B-Chat": {
+            "hf": "THUDM/chatglm2-6b",
+            "ms": "ZhipuAI/chatglm2-6b",
+        }
     },
     module="query_key_value",
     template="chatglm2"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "ChatGLM2-6B-Chat": "ZhipuAI/chatglm2-6b"
-})
-
-
 register_model_group(
     models={
-        "ChatGLM3-6B-Base": "THUDM/chatglm3-6b-base",
-        "ChatGLM3-6B-Chat": "THUDM/chatglm3-6b"
+        "ChatGLM3-6B-Base": {
+            "hf": "THUDM/chatglm3-6b-base",
+            "ms": "ZhipuAI/chatglm3-6b-base",
+        },
+        "ChatGLM3-6B-Chat": {
+            "hf": "THUDM/chatglm3-6b",
+            "ms": "ZhipuAI/chatglm3-6b",
+        }
     },
     module="query_key_value",
     template="chatglm3"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "ChatGLM3-6B-Base": "ZhipuAI/chatglm3-6b-base",
-    "ChatGLM3-6B-Chat": "ZhipuAI/chatglm3-6b"
-})
-
-
 register_model_group(
     models={
-        "ChineseLLaMA2-1.3B": "hfl/chinese-llama-2-1.3b",
-        "ChineseLLaMA2-7B": "hfl/chinese-llama-2-7b",
-        "ChineseLLaMA2-13B": "hfl/chinese-llama-2-13b",
-        "ChineseLLaMA2-1.3B-Chat": "hfl/chinese-alpaca-2-1.3b",
-        "ChineseLLaMA2-7B-Chat": "hfl/chinese-alpaca-2-7b",
-        "ChineseLLaMA2-13B-Chat": "hfl/chinese-alpaca-2-13b"
+        "ChineseLLaMA2-1.3B": {
+            "hf": "hfl/chinese-llama-2-1.3b",
+            "ms": "hfl/chinese-llama-2-1.3b",
+        },
+        "ChineseLLaMA2-7B": {
+            "hf": "hfl/chinese-llama-2-7b",
+            "ms": "hfl/chinese-llama-2-7b",
+        },
+        "ChineseLLaMA2-13B": {
+            "hf": "hfl/chinese-llama-2-13b",
+            "ms": "hfl/chinese-llama-2-13b",
+        },
+        "ChineseLLaMA2-1.3B-Chat": {
+            "hf": "hfl/chinese-alpaca-2-1.3b",
+            "ms": "hfl/chinese-alpaca-2-1.3b",
+        },
+        "ChineseLLaMA2-7B-Chat": {
+            "hf": "hfl/chinese-alpaca-2-7b",
+            "ms": "hfl/chinese-alpaca-2-7b",
+        },
+        "ChineseLLaMA2-13B-Chat": {
+            "hf": "hfl/chinese-alpaca-2-13b",
+            "ms": "hfl/chinese-alpaca-2-13b",
+        }
     },
     template="llama2_zh"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "ChineseLLaMA2-1.3B": "hfl/chinese-llama-2-1.3b",
-    "ChineseLLaMA2-7B": "hfl/chinese-llama-2-7b",
-    "ChineseLLaMA2-13B": "hfl/chinese-llama-2-13b",
-    "ChineseLLaMA2-1.3B-Chat": "hfl/chinese-alpaca-2-1.3b",
-    "ChineseLLaMA2-7B-Chat": "hfl/chinese-alpaca-2-7b",
-    "ChineseLLaMA2-13B-Chat": "hfl/chinese-alpaca-2-13b"
-})
-
-
 register_model_group(
     models={
-        "Falcon-7B": "tiiuae/falcon-7b",
-        "Falcon-40B": "tiiuae/falcon-40b",
-        "Falcon-180B": "tiiuae/falcon-180B",
-        "Falcon-7B-Chat": "tiiuae/falcon-7b-instruct",
-        "Falcon-40B-Chat": "tiiuae/falcon-40b-instruct",
-        "Falcon-180B-Chat": "tiiuae/falcon-180B-chat"
+        "Falcon-7B": {
+            "hf": "tiiuae/falcon-7b",
+            "ms": "tiiuae/falcon-7b",
+        },
+        "Falcon-40B": {
+            "hf": "tiiuae/falcon-40b",
+            "ms": "tiiuae/falcon-40b",
+        },
+        "Falcon-180B": {
+            "hf": "tiiuae/falcon-180B",
+            "ms": "tiiuae/falcon-180B",
+        },
+        "Falcon-7B-Chat": {
+            "hf": "tiiuae/falcon-7b-instruct",
+            "ms": "AI-ModelScope/falcon-7b-instruct",
+        },
+        "Falcon-40B-Chat": {
+            "hf": "tiiuae/falcon-40b-instruct",
+            "ms": "tiiuae/falcon-40b-instruct",
+        },
+        "Falcon-180B-Chat": {
+            "hf": "tiiuae/falcon-180B-chat",
+            "ms": "tiiuae/falcon-180B-chat",
+        }
     },
     module="query_key_value",
     template="falcon"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "Falcon-7B": "tiiuae/falcon-7b",
-    "Falcon-40B": "tiiuae/falcon-40b",
-    "Falcon-180B": "tiiuae/falcon-180B",
-    "Falcon-7B-Chat": "AI-ModelScope/falcon-7b-instruct",
-    "Falcon-40B-Chat": "tiiuae/falcon-40b-instruct",
-    "Falcon-180B-Chat": "tiiuae/falcon-180B-chat"
-})
-
-
 register_model_group(
     models={
-        "InternLM-7B": "internlm/internlm-7b",
-        "InternLM-20B": "internlm/internlm-20b",
-        "InternLM-7B-Chat": "internlm/internlm-chat-7b",
-        "InternLM-20B-Chat": "internlm/internlm-chat-20b"
+        "InternLM-7B": {
+            "hf": "internlm/internlm-7b",
+            "ms": "Shanghai_AI_Laboratory/internlm-7b",
+        },
+        "InternLM-20B": {
+            "hf": "internlm/internlm-20b",
+            "ms": "Shanghai_AI_Laboratory/internlm-20b",
+        },
+        "InternLM-7B-Chat": {
+            "hf": "internlm/internlm-chat-7b",
+            "ms": "Shanghai_AI_Laboratory/internlm-chat-7b",
+        },
+        "InternLM-20B-Chat": {
+            "hf": "internlm/internlm-chat-20b",
+            "ms": "Shanghai_AI_Laboratory/internlm-chat-20b",
+        }
     },
     template="intern"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "InternLM-7B": "Shanghai_AI_Laboratory/internlm-7b",
-    "InternLM-20B": "Shanghai_AI_Laboratory/internlm-20b",
-    "InternLM-7B-Chat": "Shanghai_AI_Laboratory/internlm-chat-7b",
-    "InternLM-20B-Chat": "Shanghai_AI_Laboratory/internlm-chat-20b"
-})
-
-
 register_model_group(
     models={
-        "LingoWhale-8B": "deeplang-ai/LingoWhale-8B"
+        "LingoWhale-8B": {
+            "hf": "deeplang-ai/LingoWhale-8B",
+            "ms": "DeepLang/LingoWhale-8B",
+        }
     },
     module="qkv_proj"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "LingoWhale-8B": "DeepLang/LingoWhale-8B"
-})
-
-
 register_model_group(
     models={
-        "LLaMA-7B": "huggyllama/llama-7b",
-        "LLaMA-13B": "huggyllama/llama-13b",
-        "LLaMA-30B": "huggyllama/llama-30b",
-        "LLaMA-65B": "huggyllama/llama-65b"
+        "LLaMA-7B": {
+            "hf": "huggyllama/llama-7b",
+            "ms": "skyline2006/llama-7b",
+        },
+        "LLaMA-13B": {
+            "hf": "huggyllama/llama-13b",
+            "ms": "skyline2006/llama-13b",
+        },
+        "LLaMA-30B": {
+            "hf": "huggyllama/llama-30b",
+            "ms": "skyline2006/llama-30b",
+        },
+        "LLaMA-65B": {
+            "hf": "huggyllama/llama-65b",
+            "ms": "skyline2006/llama-65b",
+        }
     }
 )
 
 
-MODELSCOPE_MODELS.update({
-    "LLaMA-7B": "skyline2006/llama-7b",
-    "LLaMA-13B": "skyline2006/llama-13b",
-    "LLaMA-30B": "skyline2006/llama-30b",
-    "LLaMA-65B": "skyline2006/llama-65b"
-})
-
-
 register_model_group(
     models={
-        "LLaMA2-7B": "meta-llama/Llama-2-7b-hf",
-        "LLaMA2-13B": "meta-llama/Llama-2-13b-hf",
-        "LLaMA2-70B": "meta-llama/Llama-2-70b-hf",
-        "LLaMA2-7B-Chat": "meta-llama/Llama-2-7b-chat-hf",
-        "LLaMA2-13B-Chat": "meta-llama/Llama-2-13b-chat-hf",
-        "LLaMA2-70B-Chat": "meta-llama/Llama-2-70b-chat-hf"
+        "LLaMA2-7B": {
+            "hf": "meta-llama/Llama-2-7b-hf",
+            "ms": "modelscope/Llama-2-7b-ms",
+        },
+        "LLaMA2-13B": {
+            "hf": "meta-llama/Llama-2-13b-hf",
+            "ms": "modelscope/Llama-2-13b-ms",
+        },
+        "LLaMA2-70B": {
+            "hf": "meta-llama/Llama-2-70b-hf",
+            "ms": "modelscope/Llama-2-70b-ms",
+        },
+        "LLaMA2-7B-Chat": {
+            "hf": "meta-llama/Llama-2-7b-chat-hf",
+            "ms": "modelscope/Llama-2-7b-chat-ms",
+        },
+        "LLaMA2-13B-Chat": {
+            "hf": "meta-llama/Llama-2-13b-chat-hf",
+            "ms": "modelscope/Llama-2-13b-chat-ms",
+        },
+        "LLaMA2-70B-Chat": {
+            "hf": "meta-llama/Llama-2-70b-chat-hf",
+            "ms": "modelscope/Llama-2-70b-chat-ms",
+        }
     },
     template="llama2"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "LLaMA2-7B": "modelscope/Llama-2-7b-ms",
-    "LLaMA2-13B": "modelscope/Llama-2-13b-ms",
-    "LLaMA2-70B": "modelscope/Llama-2-70b-ms",
-    "LLaMA2-7B-Chat": "modelscope/Llama-2-7b-chat-ms",
-    "LLaMA2-13B-Chat": "modelscope/Llama-2-13b-chat-ms",
-    "LLaMA2-70B-Chat": "modelscope/Llama-2-70b-chat-ms"
-})
-
-
 register_model_group(
     models={
-        "Mistral-7B": "mistralai/Mistral-7B-v0.1",
-        "Mistral-7B-Chat": "mistralai/Mistral-7B-Instruct-v0.1"
+        "Mistral-7B": {
+            "hf": "mistralai/Mistral-7B-v0.1",
+            "ms": "AI-ModelScope/Mistral-7B-v0.1",
+        },
+        "Mistral-7B-Chat": {
+            "hf": "mistralai/Mistral-7B-Instruct-v0.1",
+            "ms": "AI-ModelScope/Mistral-7B-Instruct-v0.1",
+        }
     },
     template="mistral"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "Mistral-7B": "AI-ModelScope/Mistral-7B-v0.1",
-    "Mistral-7B-Chat": "AI-ModelScope/Mistral-7B-Instruct-v0.1"
-})
-
-
 register_model_group(
     models={
-        "OpenChat3.5-7B-Chat": "openchat/openchat_3.5"
+        "OpenChat3.5-7B-Chat": {
+            "hf": "openchat/openchat_3.5",
+            "ms": "openchat/openchat_3.5",
+        }
     },
     template="openchat"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "OpenChat3.5-7B-Chat": "openchat/openchat_3.5"
-})
-
-
 register_model_group(
     models={
-        "Phi1.5-1.3B": "microsoft/phi-1_5"
+        "Phi1.5-1.3B": {
+            "hf": "microsoft/phi-1_5",
+            "ms": "microsoft/phi-1_5",
+        }
     },
     module="Wqkv"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "Phi1.5-1.3B": "microsoft/phi-1_5"
-})
-
-
 register_model_group(
     models={
-        "Qwen-7B": "Qwen/Qwen-7B",
-        "Qwen-14B": "Qwen/Qwen-14B",
-        "Qwen-7B-Chat": "Qwen/Qwen-7B-Chat",
-        "Qwen-14B-Chat": "Qwen/Qwen-14B-Chat",
-        "Qwen-7B-int8-Chat": "Qwen/Qwen-7B-Chat-Int8",
-        "Qwen-7B-int4-Chat": "Qwen/Qwen-7B-Chat-Int4",
-        "Qwen-14B-int8-Chat": "Qwen/Qwen-14B-Chat-Int8",
-        "Qwen-14B-int4-Chat": "Qwen/Qwen-14B-Chat-Int4"
+        "Qwen-7B": {
+            "hf": "Qwen/Qwen-7B",
+            "ms": "qwen/Qwen-7B",
+        },
+        "Qwen-14B": {
+            "hf": "Qwen/Qwen-14B",
+            "ms": "qwen/Qwen-14B",
+        },
+        "Qwen-7B-Chat": {
+            "hf": "Qwen/Qwen-7B-Chat",
+            "ms": "qwen/Qwen-7B-Chat",
+        },
+        "Qwen-14B-Chat": {
+            "hf": "Qwen/Qwen-14B-Chat",
+            "ms": "qwen/Qwen-14B-Chat",
+        },
+        "Qwen-7B-int8-Chat": {
+            "hf": "Qwen/Qwen-7B-Chat-Int8",
+            "ms": "qwen/Qwen-7B-Chat-Int8",
+        },
+        "Qwen-7B-int4-Chat": {
+            "hf": "Qwen/Qwen-7B-Chat-Int4",
+            "ms": "qwen/Qwen-7B-Chat-Int4",
+        },
+        "Qwen-14B-int8-Chat": {
+            "hf": "Qwen/Qwen-14B-Chat-Int8",
+            "ms": "qwen/Qwen-14B-Chat-Int8",
+        },
+        "Qwen-14B-int4-Chat": {
+            "hf": "Qwen/Qwen-14B-Chat-Int4",
+            "ms": "qwen/Qwen-14B-Chat-Int4",
+        }
     },
     module="c_attn",
     template="qwen"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "Qwen-7B": "qwen/Qwen-7B",
-    "Qwen-14B": "qwen/Qwen-14B",
-    "Qwen-7B-Chat": "qwen/Qwen-7B-Chat",
-    "Qwen-14B-Chat": "qwen/Qwen-14B-Chat",
-    "Qwen-7B-int8-Chat": "qwen/Qwen-7B-Chat-Int8",
-    "Qwen-7B-int4-Chat": "qwen/Qwen-7B-Chat-Int4",
-    "Qwen-14B-int8-Chat": "qwen/Qwen-14B-Chat-Int8",
-    "Qwen-14B-int4-Chat": "qwen/Qwen-14B-Chat-Int4"
-})
-
-
 register_model_group(
     models={
-        "Skywork-13B-Base": "Skywork/Skywork-13B-base"
+        "Skywork-13B-Base": {
+            "hf": "Skywork/Skywork-13B-base",
+            "ms": "skywork/Skywork-13B-base",
+        }
     }
 )
 
 
-MODELSCOPE_MODELS.update({
-    "Skywork-13B-Base": "skywork/Skywork-13B-base"
-})
-
-
 register_model_group(
     models={
-        "Vicuna1.5-7B-Chat": "lmsys/vicuna-7b-v1.5",
-        "Vicuna1.5-13B-Chat": "lmsys/vicuna-13b-v1.5"
+        "Vicuna1.5-7B-Chat": {
+            "hf": "lmsys/vicuna-7b-v1.5",
+            "ms": "AI-ModelScope/vicuna-7b-v1.5",
+        },
+        "Vicuna1.5-13B-Chat": {
+            "hf": "lmsys/vicuna-13b-v1.5",
+            "ms": "lmsys/vicuna-13b-v1.5",
+        }
     },
     template="vicuna"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "Vicuna1.5-7B-Chat": "AI-ModelScope/vicuna-7b-v1.5",
-    "Vicuna1.5-13B-Chat": "lmsys/vicuna-13b-v1.5"
-})
-
-
 register_model_group(
     models={
-        "XVERSE-7B": "xverse/XVERSE-7B",
-        "XVERSE-13B": "xverse/XVERSE-13B",
-        "XVERSE-65B": "xverse/XVERSE-65B",
-        "XVERSE-7B-Chat": "xverse/XVERSE-7B-Chat",
-        "XVERSE-13B-Chat": "xverse/XVERSE-13B-Chat"
+        "XVERSE-7B": {
+            "hf": "xverse/XVERSE-7B",
+            "ms": "xverse/XVERSE-7B",
+        },
+        "XVERSE-13B": {
+            "hf": "xverse/XVERSE-13B",
+            "ms": "xverse/XVERSE-13B",
+        },
+        "XVERSE-65B": {
+            "hf": "xverse/XVERSE-65B",
+            "ms": "xverse/XVERSE-65B",
+        },
+        "XVERSE-7B-Chat": {
+            "hf": "xverse/XVERSE-7B-Chat",
+            "ms": "xverse/XVERSE-7B-Chat",
+        },
+        "XVERSE-13B-Chat": {
+            "hf": "xverse/XVERSE-13B-Chat",
+            "ms": "xverse/XVERSE-13B-Chat",
+        }
     },
     template="xverse"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "XVERSE-7B": "xverse/XVERSE-7B",
-    "XVERSE-13B": "xverse/XVERSE-13B",
-    "XVERSE-65B": "xverse/XVERSE-65B",
-    "XVERSE-7B-Chat": "xverse/XVERSE-7B-Chat",
-    "XVERSE-13B-Chat": "xverse/XVERSE-13B-Chat"
-})
-
-
 register_model_group(
     models={
-        "Yayi-7B": "wenge-research/yayi-7b-llama2",
-        "Yayi-13B": "wenge-research/yayi-13b-llama2"
+        "Yayi-7B": {
+            "hf": "wenge-research/yayi-7b-llama2",
+            "ms": "wenge-research/yayi-7b-llama2",
+        },
+        "Yayi-13B": {
+            "hf": "wenge-research/yayi-13b-llama2",
+            "ms": "wenge-research/yayi-13b-llama2",
+        }
     },
     template="yayi"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "Yayi-7B": "wenge-research/yayi-7b-llama2",
-    "Yayi-13B": "wenge-research/yayi-13b-llama2"
-})
-
-
 register_model_group(
     models={
-        "Yi-6B": "01-ai/Yi-6B",
-        "Yi-34B": "01-ai/Yi-34B",
-        "Yi-34B-Chat": "01-ai/Yi-34B-Chat",
-        "Yi-34B-int8-Chat": "01-ai/Yi-34B-Chat-8bits"
+        "Yi-6B": {
+            "hf": "01-ai/Yi-6B",
+            "ms": "01ai/Yi-6B",
+        },
+        "Yi-34B": {
+            "hf": "01-ai/Yi-34B",
+            "ms": "01ai/Yi-34B",
+        },
+        "Yi-34B-Chat": {
+            "hf": "01-ai/Yi-34B-Chat",
+            "ms": "01ai/Yi-34B-Chat",
+        },
+        "Yi-34B-int8-Chat": {
+            "hf": "01-ai/Yi-34B-Chat-8bits",
+            "ms": "01ai/Yi-34B-Chat-8bits",
+        }
     },
     template="yi"
 )
 
 
-MODELSCOPE_MODELS.update({
-    "Yi-6B": "01ai/Yi-6B",
-    "Yi-34B": "01ai/Yi-34B",
-    "Yi-34B-Chat": "01ai/Yi-34B-Chat",
-    "Yi-34B-int8-Chat": "01ai/Yi-34B-Chat-8bits"
-})
-
-
 register_model_group(
     models={
-        "Zephyr-7B-Alpha-Chat": "HuggingFaceH4/zephyr-7b-alpha",
-        "Zephyr-7B-Beta-Chat": "HuggingFaceH4/zephyr-7b-beta"
+        "Zephyr-7B-Alpha-Chat": {
+            "hf": "HuggingFaceH4/zephyr-7b-alpha",
+            "ms": "HuggingFaceH4/zephyr-7b-alpha",
+        },
+        "Zephyr-7B-Beta-Chat": {
+            "hf": "HuggingFaceH4/zephyr-7b-beta",
+            "ms": "modelscope/zephyr-7b-beta",
+        }
     },
     template="zephyr"
 )
-
-
-MODELSCOPE_MODELS.update({
-    "Zephyr-7B-Alpha-Chat": "HuggingFaceH4/zephyr-7b-alpha",
-    "Zephyr-7B-Beta-Chat": "modelscope/zephyr-7b-beta"
-})
diff --git a/src/llmtuner/model/loader.py b/src/llmtuner/model/loader.py
index 5fed514a..85e7ed31 100644
--- a/src/llmtuner/model/loader.py
+++ b/src/llmtuner/model/loader.py
@@ -238,7 +238,10 @@ def try_download_model_from_ms(model_args):
     if os.environ.get('USE_MODELSCOPE_HUB', False) and not os.path.exists(model_args.model_name_or_path):
         try:
             from modelscope import snapshot_download
-            model_args.model_name_or_path = snapshot_download(model_args.model_name_or_path, model_args.model_revision)
+            revision = model_args.model_revision
+            if revision == 'main':
+                revision = 'master'
+            model_args.model_name_or_path = snapshot_download(model_args.model_name_or_path, revision)
         except ImportError as e:
             raise ImportError(f'You are using `USE_MODELSCOPE_HUB=True` but you have no modelscope sdk installed. '
                               f'Please install it by `pip install modelscope -U`') from e

From 63e12226a0d3f215eef5ab4655af1c05a4bafca8 Mon Sep 17 00:00:00 2001
From: tastelikefeet <yuze.zyz@alibaba-inc.com>
Date: Fri, 1 Dec 2023 15:06:17 +0800
Subject: [PATCH 3/6] add model

Former-commit-id: 8ce4d11e38518b0b4657c7e64394d471cbb0bd6d
---
 src/llmtuner/extras/constants.py | 42 ++++++++++++++++----------------
 1 file changed, 21 insertions(+), 21 deletions(-)

diff --git a/src/llmtuner/extras/constants.py b/src/llmtuner/extras/constants.py
index 9de07bc2..3930a2e3 100644
--- a/src/llmtuner/extras/constants.py
+++ b/src/llmtuner/extras/constants.py
@@ -107,11 +107,11 @@ register_model_group(
         },
         "BLOOM-3B": {
             "hf": "bigscience/bloom-3b",
-            "ms": "bigscience/bloom-3b",
+            "ms": "AI-ModelScope/bloom-3b",
         },
         "BLOOM-7B1": {
             "hf": "bigscience/bloom-7b1",
-            "ms": "bigscience/bloom-7b1",
+            "ms": "AI-ModelScope/bloom-7b1",
         }
     },
     module="query_key_value"
@@ -122,11 +122,11 @@ register_model_group(
     models={
         "BLOOMZ-560M": {
             "hf": "bigscience/bloomz-560m",
-            "ms": "bigscience/bloomz-560m",
+            "ms": "AI-ModelScope/bloomz-560m",
         },
         "BLOOMZ-3B": {
             "hf": "bigscience/bloomz-3b",
-            "ms": "bigscience/bloomz-3b",
+            "ms": "AI-ModelScope/bloomz-3b",
         },
         "BLOOMZ-7B1-mt": {
             "hf": "bigscience/bloomz-7b1-mt",
@@ -184,27 +184,27 @@ register_model_group(
     models={
         "ChineseLLaMA2-1.3B": {
             "hf": "hfl/chinese-llama-2-1.3b",
-            "ms": "hfl/chinese-llama-2-1.3b",
+            "ms": "AI-ModelScope/chinese-llama-2-1.3b",
         },
         "ChineseLLaMA2-7B": {
             "hf": "hfl/chinese-llama-2-7b",
-            "ms": "hfl/chinese-llama-2-7b",
+            "ms": "AI-ModelScope/chinese-llama-2-7b",
         },
         "ChineseLLaMA2-13B": {
             "hf": "hfl/chinese-llama-2-13b",
-            "ms": "hfl/chinese-llama-2-13b",
+            "ms": "AI-ModelScope/chinese-llama-2-13b",
         },
         "ChineseLLaMA2-1.3B-Chat": {
             "hf": "hfl/chinese-alpaca-2-1.3b",
-            "ms": "hfl/chinese-alpaca-2-1.3b",
+            "ms": "AI-ModelScope/chinese-alpaca-2-1.3b",
         },
         "ChineseLLaMA2-7B-Chat": {
             "hf": "hfl/chinese-alpaca-2-7b",
-            "ms": "hfl/chinese-alpaca-2-7b",
+            "ms": "AI-ModelScope/chinese-alpaca-2-7b",
         },
         "ChineseLLaMA2-13B-Chat": {
             "hf": "hfl/chinese-alpaca-2-13b",
-            "ms": "hfl/chinese-alpaca-2-13b",
+            "ms": "AI-ModelScope/chinese-alpaca-2-13b",
         }
     },
     template="llama2_zh"
@@ -215,15 +215,15 @@ register_model_group(
     models={
         "Falcon-7B": {
             "hf": "tiiuae/falcon-7b",
-            "ms": "tiiuae/falcon-7b",
+            "ms": "AI-ModelScope/falcon-7b",
         },
         "Falcon-40B": {
             "hf": "tiiuae/falcon-40b",
-            "ms": "tiiuae/falcon-40b",
+            "ms": "AI-ModelScope/falcon-40b",
         },
         "Falcon-180B": {
             "hf": "tiiuae/falcon-180B",
-            "ms": "tiiuae/falcon-180B",
+            "ms": "AI-ModelScope/falcon-180B",
         },
         "Falcon-7B-Chat": {
             "hf": "tiiuae/falcon-7b-instruct",
@@ -231,11 +231,11 @@ register_model_group(
         },
         "Falcon-40B-Chat": {
             "hf": "tiiuae/falcon-40b-instruct",
-            "ms": "tiiuae/falcon-40b-instruct",
+            "ms": "AI-ModelScope/falcon-40b-instruct",
         },
         "Falcon-180B-Chat": {
             "hf": "tiiuae/falcon-180B-chat",
-            "ms": "tiiuae/falcon-180B-chat",
+            "ms": "AI-ModelScope/falcon-180B-chat",
         }
     },
     module="query_key_value",
@@ -349,7 +349,7 @@ register_model_group(
     models={
         "OpenChat3.5-7B-Chat": {
             "hf": "openchat/openchat_3.5",
-            "ms": "openchat/openchat_3.5",
+            "ms": "myxiongmodel/openchat_3.5",
         }
     },
     template="openchat"
@@ -360,7 +360,7 @@ register_model_group(
     models={
         "Phi1.5-1.3B": {
             "hf": "microsoft/phi-1_5",
-            "ms": "microsoft/phi-1_5",
+            "ms": "allspace/PHI_1-5",
         }
     },
     module="Wqkv"
@@ -425,7 +425,7 @@ register_model_group(
         },
         "Vicuna1.5-13B-Chat": {
             "hf": "lmsys/vicuna-13b-v1.5",
-            "ms": "lmsys/vicuna-13b-v1.5",
+            "ms": "Xorbits/vicuna-13b-v1.5",
         }
     },
     template="vicuna"
@@ -463,11 +463,11 @@ register_model_group(
     models={
         "Yayi-7B": {
             "hf": "wenge-research/yayi-7b-llama2",
-            "ms": "wenge-research/yayi-7b-llama2",
+            "ms": "AI-ModelScope/yayi-7b-llama2",
         },
         "Yayi-13B": {
             "hf": "wenge-research/yayi-13b-llama2",
-            "ms": "wenge-research/yayi-13b-llama2",
+            "ms": "AI-ModelScope/yayi-13b-llama2",
         }
     },
     template="yayi"
@@ -501,7 +501,7 @@ register_model_group(
     models={
         "Zephyr-7B-Alpha-Chat": {
             "hf": "HuggingFaceH4/zephyr-7b-alpha",
-            "ms": "HuggingFaceH4/zephyr-7b-alpha",
+            "ms": "AI-ModelScope/zephyr-7b-alpha",
         },
         "Zephyr-7B-Beta-Chat": {
             "hf": "HuggingFaceH4/zephyr-7b-beta",

From b2200409f57d1a67237b5be1cc27c29b1a5b57b7 Mon Sep 17 00:00:00 2001
From: "yuze.zyz" <yuze.zyz@alibaba-inc.com>
Date: Fri, 1 Dec 2023 16:11:30 +0800
Subject: [PATCH 4/6] add readme

Former-commit-id: 5aa6751e52b5c2e06727c50e60218226b146b7bf
---
 README.md                        | 29 +++++++++++++++++++++++++++++
 README_zh.md                     | 29 +++++++++++++++++++++++++++++
 src/llmtuner/extras/constants.py |  2 +-
 src/llmtuner/model/loader.py     |  4 ++--
 4 files changed, 61 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index 7ef3b6f1..0a03d61c 100644
--- a/README.md
+++ b/README.md
@@ -51,6 +51,8 @@ Compared to ChatGLM's [P-Tuning](https://github.com/THUDM/ChatGLM2-6B/tree/main/
 
 ## Changelog
 
+[23/12/01] We supported **[ModelScope Hub](https://www.modelscope.cn/models)** to accelerate model downloading. Add environment variable `USE_MODELSCOPE_HUB=1` to your command line, then you can use the model-id of ModelScope Hub.
+
 [23/10/21] We supported **[NEFTune](https://arxiv.org/abs/2310.05914)** trick for fine-tuning. Try `--neft_alpha` argument to activate NEFTune, e.g., `--neft_alpha 5`.
 
 [23/09/27] We supported **$S^2$-Attn** proposed by [LongLoRA](https://github.com/dvlab-research/LongLoRA) for the LLaMA models. Try `--shift_attn` argument to enable shift short attention.
@@ -227,6 +229,33 @@ If you want to enable the quantized LoRA (QLoRA) on the Windows platform, you wi
 pip install https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.39.1-py3-none-win_amd64.whl
 ```
 
+### Use ModelScope Models
+
+If you have trouble with downloading models from HuggingFace, we have supported ModelScope Hub. To use LLaMA-Factory together with ModelScope, please add a environment variable:
+
+```shell
+export USE_MODELSCOPE_HUB=1
+```
+
+> [!NOTE]
+>
+> Please use integers only. 0 or not set for using HuggingFace hub. Other values will be treated as use ModelScope hub.
+
+Then you can use LLaMA-Factory with ModelScope model-ids:
+
+```shell
+python src/train_bash.py \
+    --model_name_or_path ZhipuAI/chatglm3-6b \
+    ... other arguments
+# You can find all model ids in this link: https://www.modelscope.cn/models
+```
+
+Web demo also supports ModelScope, after setting the environment variable please run with this command:
+
+```shell
+CUDA_VISIBLE_DEVICES=0 python src/train_web.py
+```
+
 ### Train on a single GPU
 
 > [!IMPORTANT]
diff --git a/README_zh.md b/README_zh.md
index d6f01d2f..9c92c261 100644
--- a/README_zh.md
+++ b/README_zh.md
@@ -51,6 +51,8 @@ https://github.com/hiyouga/LLaMA-Factory/assets/16256802/6ba60acc-e2e2-4bec-b846
 
 ## 更新日志
 
+[23/12/01] 我们支持了 **[魔搭ModelHub](https://www.modelscope.cn/models)** 进行模型下载加速。在启动命令前环境变量中增加 `USE_MODELSCOPE_HUB=1` 即可开启。
+
 [23/10/21] 我们支持了 **[NEFTune](https://arxiv.org/abs/2310.05914)** 训练技巧。请使用 `--neft_alpha` 参数启用 NEFTune，例如 `--neft_alpha 5`。
 
 [23/09/27] 我们针对 LLaMA 模型支持了 [LongLoRA](https://github.com/dvlab-research/LongLoRA) 提出的 **$S^2$-Attn**。请使用 `--shift_attn` 参数以启用该功能。
@@ -227,6 +229,33 @@ pip install -r requirements.txt
 pip install https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.39.1-py3-none-win_amd64.whl
 ```
 
+### 使用魔搭的模型
+
+如果下载HuggingFace模型存在问题，我们已经支持了魔搭的ModelHub，只需要添加一个环境变量：
+
+```shell
+export USE_MODELSCOPE_HUB=1
+```
+
+> [!NOTE]
+>
+> 该环境变量仅支持整数，0或者不设置代表使用HuggingFace，其他值代表使用ModelScope
+
+之后就可以在命令行中指定魔搭的模型id：
+
+```shell
+python src/train_bash.py \
+    --model_name_or_path ZhipuAI/chatglm3-6b \
+    ... other arguments
+# 在这个链接中可以看到所有可用模型: https://www.modelscope.cn/models
+```
+
+Web demo目前也支持了魔搭, 在设置环境变量后即可使用：
+
+```shell
+CUDA_VISIBLE_DEVICES=0 python src/train_web.py
+```
+
 ### 单 GPU 训练
 
 > [!IMPORTANT]
diff --git a/src/llmtuner/extras/constants.py b/src/llmtuner/extras/constants.py
index 3930a2e3..c865b4f8 100644
--- a/src/llmtuner/extras/constants.py
+++ b/src/llmtuner/extras/constants.py
@@ -43,7 +43,7 @@ def register_model_group(
         else:
             assert prefix == name.split("-")[0], "prefix should be identical."
 
-        if not os.environ.get('USE_MODELSCOPE_HUB', False):
+        if not int(os.environ.get('USE_MODELSCOPE_HUB', '0')):
             # If path is a string, we treat it as a huggingface model-id by default.
             SUPPORTED_MODELS[name] = path["hf"] if isinstance(path, dict) else path
         elif isinstance(path, dict) and "ms" in path:
diff --git a/src/llmtuner/model/loader.py b/src/llmtuner/model/loader.py
index 85e7ed31..122cd7f2 100644
--- a/src/llmtuner/model/loader.py
+++ b/src/llmtuner/model/loader.py
@@ -235,7 +235,7 @@ def load_model_and_tokenizer(
 
 
 def try_download_model_from_ms(model_args):
-    if os.environ.get('USE_MODELSCOPE_HUB', False) and not os.path.exists(model_args.model_name_or_path):
+    if int(os.environ.get('USE_MODELSCOPE_HUB', '0')) and not os.path.exists(model_args.model_name_or_path):
         try:
             from modelscope import snapshot_download
             revision = model_args.model_revision
@@ -243,5 +243,5 @@ def try_download_model_from_ms(model_args):
                 revision = 'master'
             model_args.model_name_or_path = snapshot_download(model_args.model_name_or_path, revision)
         except ImportError as e:
-            raise ImportError(f'You are using `USE_MODELSCOPE_HUB=True` but you have no modelscope sdk installed. '
+            raise ImportError(f'You are using `USE_MODELSCOPE_HUB=1` but you have no modelscope sdk installed. '
                               f'Please install it by `pip install modelscope -U`') from e

From eb835b693d22c154d986ba5ff31e5f878e2c00e8 Mon Sep 17 00:00:00 2001
From: tastelikefeet <yuze.zyz@alibaba-inc.com>
Date: Fri, 1 Dec 2023 17:27:00 +0800
Subject: [PATCH 5/6] fix bug

Former-commit-id: d9e52957e272e8133f1b37cf20d193084425e09e
---
 src/llmtuner/extras/constants.py | 4 +++-
 src/llmtuner/webui/common.py     | 7 +++++--
 2 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/src/llmtuner/extras/constants.py b/src/llmtuner/extras/constants.py
index c865b4f8..69f4510d 100644
--- a/src/llmtuner/extras/constants.py
+++ b/src/llmtuner/extras/constants.py
@@ -20,7 +20,7 @@ SUBJECTS = ["Average", "STEM", "Social Sciences", "Humanities", "Other"]
 
 SUPPORTED_MODELS = OrderedDict()
 
-MODELSCOPE_MODELS = OrderedDict()
+ALL_OFFICIAL_MODELS = OrderedDict()
 
 TRAINING_STAGES = {
     "Supervised Fine-Tuning": "sft",
@@ -43,12 +43,14 @@ def register_model_group(
         else:
             assert prefix == name.split("-")[0], "prefix should be identical."
 
+        ALL_OFFICIAL_MODELS[name] = [path] if isinstance(path, str) else list(path.values())
         if not int(os.environ.get('USE_MODELSCOPE_HUB', '0')):
             # If path is a string, we treat it as a huggingface model-id by default.
             SUPPORTED_MODELS[name] = path["hf"] if isinstance(path, dict) else path
         elif isinstance(path, dict) and "ms" in path:
             # Use ModelScope modelhub
             SUPPORTED_MODELS[name] = path["ms"]
+        print(f'Supported models add {name}/{SUPPORTED_MODELS[name]}')
     if module is not None:
         DEFAULT_MODULE[prefix] = module
     if template is not None:
diff --git a/src/llmtuner/webui/common.py b/src/llmtuner/webui/common.py
index 55d8942b..b21cad62 100644
--- a/src/llmtuner/webui/common.py
+++ b/src/llmtuner/webui/common.py
@@ -11,7 +11,7 @@ from transformers.utils import (
     ADAPTER_SAFE_WEIGHTS_NAME
 )
 
-from llmtuner.extras.constants import DEFAULT_MODULE, DEFAULT_TEMPLATE, SUPPORTED_MODELS, TRAINING_STAGES
+from llmtuner.extras.constants import DEFAULT_MODULE, DEFAULT_TEMPLATE, SUPPORTED_MODELS, ALL_OFFICIAL_MODELS, TRAINING_STAGES
 
 
 DEFAULT_CACHE_DIR = "cache"
@@ -58,7 +58,10 @@ def save_config(lang: str, model_name: Optional[str] = None, model_path: Optiona
 
 def get_model_path(model_name: str) -> str:
     user_config = load_config()
-    return user_config["path_dict"].get(model_name, None) or SUPPORTED_MODELS.get(model_name, "")
+    cached_path = user_config["path_dict"].get(model_name, None)
+    if cached_path in ALL_OFFICIAL_MODELS.get(model_name, []):
+        cached_path = None
+    return cached_path or SUPPORTED_MODELS.get(model_name, "")
 
 
 def get_prefix(model_name: str) -> str:

From fcd61657ee5a7591f88e0211cdc587b99f90bfb2 Mon Sep 17 00:00:00 2001
From: "yuze.zyz" <yuze.zyz@alibaba-inc.com>
Date: Fri, 1 Dec 2023 17:28:23 +0800
Subject: [PATCH 6/6] remove useless code

Former-commit-id: 5a2392f105704810e9ce96c13fcc8a555726f9b8
---
 src/llmtuner/extras/constants.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/src/llmtuner/extras/constants.py b/src/llmtuner/extras/constants.py
index 69f4510d..7e66d1b3 100644
--- a/src/llmtuner/extras/constants.py
+++ b/src/llmtuner/extras/constants.py
@@ -50,7 +50,6 @@ def register_model_group(
         elif isinstance(path, dict) and "ms" in path:
             # Use ModelScope modelhub
             SUPPORTED_MODELS[name] = path["ms"]
-        print(f'Supported models add {name}/{SUPPORTED_MODELS[name]}')
     if module is not None:
         DEFAULT_MODULE[prefix] = module
     if template is not None: