updata to v0.9.1_stable

data/mllm_demo.json

@@ -137,4 +137,4 @@
       "mllm_demo_data/3.jpg"
     ]
   }
-]
\ No newline at end of file
+]

data/mllm_video_demo.json

@@ -44,4 +44,4 @@
       "mllm_demo_data/3.mp4"
     ]
   }
-]
\ No newline at end of file
+]

data/ultra_chat/ultra_chat.py

@@ -20,9 +20,9 @@ _CITATION = """\
 }
 """
 
-_HOMEPAGE = "{}/datasets/stingning/ultrachat".format(_HF_ENDPOINT)
+_HOMEPAGE = f"{_HF_ENDPOINT}/datasets/stingning/ultrachat"
 _LICENSE = "cc-by-nc-4.0"
-_BASE_DATA_URL = "{}/datasets/stingning/ultrachat/resolve/main/train_{{idx}}.jsonl".format(_HF_ENDPOINT)
+_BASE_DATA_URL = f"{_HF_ENDPOINT}/datasets/stingning/ultrachat/resolve/main/train_{{idx}}.jsonl"
 
 
 class UltraChat(datasets.GeneratorBasedBuilder):
@@ -42,7 +42,7 @@ class UltraChat(datasets.GeneratorBasedBuilder):
 
     def _generate_examples(self, filepaths: List[str]):
         for filepath in filepaths:
-            with open(filepath, "r", encoding="utf-8") as f:
+            with open(filepath, encoding="utf-8") as f:
                 for row in f:
                     try:
                         data = json.loads(row)

docker/docker-cuda/Dockerfile

-# Use the NVIDIA official image with PyTorch 2.3.0
-# https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes/rel-24-02.html
-FROM nvcr.io/nvidia/pytorch:24.02-py3
+# Default use the NVIDIA official image with PyTorch 2.3.0
+# https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes/index.html
+ARG BASE_IMAGE=nvcr.io/nvidia/pytorch:24.02-py3
+FROM ${BASE_IMAGE}
 
 # Define environments
 ENV MAX_JOBS=4

docker/docker-cuda/docker-compose.yml

@@ -16,6 +16,7 @@ services:
     volumes:
       - ../../hf_cache:/root/.cache/huggingface
       - ../../ms_cache:/root/.cache/modelscope
+      - ../../om_cache:/root/.cache/openmind
       - ../../data:/app/data
       - ../../output:/app/output
     ports:
@@ -23,6 +24,7 @@ services:
       - "8000:8000"
     ipc: host
     tty: true
+    shm_size: '16gb'
     stdin_open: true
     command: bash
     deploy:

docker/docker-npu/docker-compose.yml

@@ -10,6 +10,7 @@ services:
     volumes:
       - ../../hf_cache:/root/.cache/huggingface
       - ../../ms_cache:/root/.cache/modelscope
+      - ../../om_cache:/root/.cache/openmind
       - ../../data:/app/data
       - ../../output:/app/output
       - /usr/local/dcmi:/usr/local/dcmi
@@ -21,6 +22,7 @@ services:
       - "8000:8000"
     ipc: host
     tty: true
+    shm_size: '16gb'
     stdin_open: true
     command: bash
     devices:

docker/docker-rocm/docker-compose.yml

@@ -15,6 +15,7 @@ services:
     volumes:
       - ../../hf_cache:/root/.cache/huggingface
       - ../../ms_cache:/root/.cache/modelscope
+      - ../../om_cache:/root/.cache/openmind
       - ../../data:/app/data
       - ../../output:/app/output
       - ../../saves:/app/saves
@@ -23,6 +24,7 @@ services:
       - "8000:8000"
     ipc: host
     tty: true
+    shm_size: '16gb'
     stdin_open: true
     command: bash
     devices:

evaluation/ceval/mapping.json

@@ -207,4 +207,4 @@
     "name": "兽医学",
     "category": "STEM"
   }
-}
\ No newline at end of file
+}

evaluation/cmmlu/mapping.json

@@ -267,4 +267,4 @@
     "name": "世界宗教",
     "category": "Humanities"
   }
-}
\ No newline at end of file
+}

evaluation/mmlu/mapping.json

@@ -227,4 +227,4 @@
     "name": "world religions",
     "category": "Humanities"
   }
-}
\ No newline at end of file
+}

evaluation/mmlu/mmlu.py

@@ -158,5 +158,4 @@ class MMLU(datasets.GeneratorBasedBuilder):
         df = pd.read_csv(filepath, header=None)
         df.columns = ["question", "A", "B", "C", "D", "answer"]
 
-        for i, instance in enumerate(df.to_dict(orient="records")):
-            yield i, instance
+        yield from enumerate(df.to_dict(orient="records"))

examples/README.md

@@ -89,8 +89,8 @@ llamafactory-cli train examples/train_lora/llama3_lora_predict.yaml
 #### Supervised Fine-Tuning on Multiple Nodes
 
 ```bash
-FORCE_TORCHRUN=1 NNODES=2 RANK=0 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
-FORCE_TORCHRUN=1 NNODES=2 RANK=1 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=0 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=1 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
 ```
 
 #### Supervised Fine-Tuning with DeepSpeed ZeRO-3 (Weight Sharding)

examples/README_zh.md

@@ -89,8 +89,8 @@ llamafactory-cli train examples/train_lora/llama3_lora_predict.yaml
 #### 多机指令监督微调
 
 ```bash
-FORCE_TORCHRUN=1 NNODES=2 RANK=0 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
-FORCE_TORCHRUN=1 NNODES=2 RANK=1 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=0 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=1 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
 ```
 
 #### 使用 DeepSpeed ZeRO-3 平均分配显存

examples/deepspeed/ds_z0_config.json

@@ -25,4 +25,4 @@
     "contiguous_gradients": true,
     "round_robin_gradients": true
   }
-}
\ No newline at end of file
+}

examples/deepspeed/ds_z2_config.json

@@ -25,4 +25,4 @@
     "contiguous_gradients": true,
     "round_robin_gradients": true
   }
-}
\ No newline at end of file
+}

examples/deepspeed/ds_z2_offload_config.json

@@ -29,4 +29,4 @@
     "contiguous_gradients": true,
     "round_robin_gradients": true
   }
-}
\ No newline at end of file
+}

examples/deepspeed/ds_z3_config.json

@@ -27,4 +27,4 @@
     "stage3_max_reuse_distance": 1e9,
     "stage3_gather_16bit_weights_on_model_save": true
   }
-}
\ No newline at end of file
+}

examples/deepspeed/ds_z3_offload_config.json

@@ -35,4 +35,4 @@
     "stage3_max_reuse_distance": 1e9,
     "stage3_gather_16bit_weights_on_model_save": true
   }
-}
\ No newline at end of file
+}

examples/extras/adam_mini/qwen2_full_sft.yaml

@@ -10,7 +10,7 @@ use_adam_mini: true
 ### dataset
 dataset: identity,alpaca_en_demo
 template: qwen
-cutoff_len: 1024
+cutoff_len: 2048
 max_samples: 1000
 overwrite_cache: true
 preprocessing_num_workers: 16