qwen/demo.py

@@ -14,7 +14,7 @@ model_dir = snapshot_download("qwen/Qwen-1_8B-Chat")
 # model_dir = "/home/colin/.cache/modelscope/hub/qwen/Qwen-1_8B-Chat"
 
 config, kwargs = AutoConfig.from_pretrained(
-    "./",
+    model_dir,
     return_unused_kwargs=True,
     trust_remote_code=True,
     code_revision=None,
@@ -25,15 +25,15 @@ model = QWenLMHeadModel(config)
 
 tokenizer = AutoTokenizer.from_pretrained(model_dir, trust_remote_code=True)
 model = model.from_pretrained(
-    model_dir, config=config, device_map="auto", trust_remote_code=True
+    model_dir, device_map="auto", trust_remote_code=True
 ).train()
 # model.train()
 # model.zero_grad()
 
 # 可指定不同的生成长度、top_p等相关超参
-# model.generation_config = GenerationConfig.from_pretrained(
-#     model_dir, trust_remote_code=True
-# )
+model.generation_config = GenerationConfig.from_pretrained(
+    model_dir, trust_remote_code=True
+)
 
 # 第一轮对话
 response, history = model.chat(tokenizer, "你好", history=None)

qwen/ds_config_zero2.json

@@ -0,0 +1,52 @@
+{
+    "fp16": {
+        "enabled": "auto",
+        "loss_scale": 0,
+        "loss_scale_window": 1000,
+        "initial_scale_power": 16,
+        "hysteresis": 2,
+        "min_loss_scale": 1
+    },
+    "bf16": {
+        "enabled": "auto"
+    },
+    "optimizer": {
+        "type": "AdamW",
+        "params": {
+            "lr": "auto",
+            "betas": "auto",
+            "eps": "auto",
+            "weight_decay": "auto"
+        }
+    },
+
+    "scheduler": {
+        "type": "WarmupLR",
+        "params": {
+            "warmup_min_lr": "auto",
+            "warmup_max_lr": "auto",
+            "warmup_num_steps": "auto"
+        }
+    },
+
+    "zero_optimization": {
+        "stage": 2,
+        "offload_optimizer": {
+            "device": "none",
+            "pin_memory": true
+        },
+        "allgather_partitions": true,
+        "allgather_bucket_size": 2e8,
+        "overlap_comm": true,
+        "reduce_scatter": true,
+        "reduce_bucket_size": 2e8,
+        "contiguous_gradients": true
+    },
+
+    "gradient_accumulation_steps": "auto",
+    "gradient_clipping": "auto",
+    "steps_per_print": 100,
+    "train_batch_size": "auto",
+    "train_micro_batch_size_per_gpu": "auto",
+    "wall_clock_breakdown": false
+}

qwen/ds_config_zero3.json

@@ -0,0 +1,59 @@
+{
+    "fp16": {
+        "enabled": "auto",
+        "loss_scale": 0,
+        "loss_scale_window": 1000,
+        "initial_scale_power": 16,
+        "hysteresis": 2,
+        "min_loss_scale": 1
+    },
+    "bf16": {
+        "enabled": "auto"
+    },
+    "optimizer": {
+        "type": "AdamW",
+        "params": {
+            "lr": "auto",
+            "betas": "auto",
+            "eps": "auto",
+            "weight_decay": "auto"
+        }
+    },
+
+    "scheduler": {
+        "type": "WarmupLR",
+        "params": {
+            "warmup_min_lr": "auto",
+            "warmup_max_lr": "auto",
+            "warmup_num_steps": "auto"
+        }
+    },
+
+    "zero_optimization": {
+        "stage": 3,
+        "offload_optimizer": {
+            "device": "none",
+            "pin_memory": true
+        },
+        "offload_param": {
+            "device": "none",
+            "pin_memory": true
+        },
+        "overlap_comm": true,
+        "contiguous_gradients": true,
+        "sub_group_size": 1e9,
+        "reduce_bucket_size": "auto",
+        "stage3_prefetch_bucket_size": "auto",
+        "stage3_param_persistence_threshold": "auto",
+        "stage3_max_live_parameters": 1e9,
+        "stage3_max_reuse_distance": 1e9,
+        "stage3_gather_16bit_weights_on_model_save": true
+    },
+
+    "gradient_accumulation_steps": "auto",
+    "gradient_clipping": "auto",
+    "steps_per_print": 100,
+    "train_batch_size": "auto",
+    "train_micro_batch_size_per_gpu": "auto",
+    "wall_clock_breakdown": false
+}

qwen/finetune_ds.sh

@@ -0,0 +1,90 @@
+#!/bin/bash
+export CUDA_DEVICE_MAX_CONNECTIONS=1
+DIR=`pwd`
+
+# Guide:
+# This script supports distributed training on multi-gpu workers (as well as single-worker training).
+# Please set the options below according to the comments.
+# For multi-gpu workers training, these options should be manually set for each worker.
+# After setting the options, please run the script on each worker.
+
+# Number of GPUs per GPU worker
+GPUS_PER_NODE=$(python -c 'import torch; print(torch.cuda.device_count())')
+
+# Number of GPU workers, for single-worker training, please set to 1
+NNODES=${NNODES:-1}
+
+# The rank of this worker, should be in {0, ..., WORKER_CNT-1}, for single-worker training, please set to 0
+NODE_RANK=${NODE_RANK:-0}
+
+# The ip address of the rank-0 worker, for single-worker training, please set to localhost
+MASTER_ADDR=${MASTER_ADDR:localhost}
+
+# The port for communication
+MASTER_PORT=${MASTER_PORT:-6001}
+
+MODEL="Qwen/Qwen-7B" # Set the path if you do not want to load from huggingface directly
+# ATTENTION: specify the path to your training data, which should be a json file consisting of a list of conversations.
+# See the section for finetuning in README for more information.
+DATA="path_to_data"
+
+function usage() {
+    echo '
+Usage: bash finetune/finetune_ds.sh [-m MODEL_PATH] [-d DATA_PATH]
+'
+}
+
+while [[ "$1" != "" ]]; do
+    case $1 in
+        -m | --model )
+            shift
+            MODEL=$1
+            ;;
+        -d | --data )
+            shift
+            DATA=$1
+            ;;
+        -h | --help )
+            usage
+            exit 0
+            ;;
+        * )
+            echo "Unknown argument ${1}"
+            exit 1
+            ;;
+    esac
+    shift
+done
+
+DISTRIBUTED_ARGS="
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --node_rank $NODE_RANK \
+    --master_addr $MASTER_ADDR \
+    --master_port $MASTER_PORT
+"
+
+torchrun $DISTRIBUTED_ARGS finetune.py \
+    --model_name_or_path $MODEL \
+    --data_path $DATA \
+    --bf16 True \
+    --output_dir output_qwen \
+    --num_train_epochs 5 \
+    --per_device_train_batch_size 1 \
+    --per_device_eval_batch_size 1 \
+    --gradient_accumulation_steps 16 \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 1000 \
+    --save_total_limit 10 \
+    --learning_rate 1e-5 \
+    --weight_decay 0.1 \
+    --adam_beta2 0.95 \
+    --warmup_ratio 0.01 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --report_to "none" \
+    --model_max_length 512 \
+    --gradient_checkpointing True \
+    --lazy_preprocess True \
+    --deepspeed finetune/ds_config_zero3.json

qwen/finetune_lora_ds.sh

@@ -0,0 +1,96 @@
+#!/bin/bash
+export CUDA_DEVICE_MAX_CONNECTIONS=1
+DIR=`pwd`
+
+# Guide:
+# This script supports distributed training on multi-gpu workers (as well as single-worker training).
+# Please set the options below according to the comments.
+# For multi-gpu workers training, these options should be manually set for each worker.
+# After setting the options, please run the script on each worker.
+
+# Number of GPUs per GPU worker
+GPUS_PER_NODE=$(python -c 'import torch; print(torch.cuda.device_count())')
+
+# Number of GPU workers, for single-worker training, please set to 1
+NNODES=${NNODES:-1}
+
+# The rank of this worker, should be in {0, ..., WORKER_CNT-1}, for single-worker training, please set to 0
+NODE_RANK=${NODE_RANK:-0}
+
+# The ip address of the rank-0 worker, for single-worker training, please set to localhost
+MASTER_ADDR=${MASTER_ADDR:localhost}
+
+# The port for communication
+MASTER_PORT=${MASTER_PORT:-6001}
+
+MODEL="Qwen/Qwen-7B" # Set the path if you do not want to load from huggingface directly
+# ATTENTION: specify the path to your training data, which should be a json file consisting of a list of conversations.
+# See the section for finetuning in README for more information.
+DATA="path_to_data"
+DS_CONFIG_PATH="finetune/ds_config_zero2.json"
+
+function usage() {
+    echo '
+Usage: bash finetune/finetune_lora_ds.sh [-m MODEL_PATH] [-d DATA_PATH] [--deepspeed DS_CONFIG_PATH]
+'
+}
+
+while [[ "$1" != "" ]]; do
+    case $1 in
+        -m | --model )
+            shift
+            MODEL=$1
+            ;;
+        -d | --data )
+            shift
+            DATA=$1
+            ;;
+        --deepspeed )
+            shift
+            DS_CONFIG_PATH=$1
+            ;;
+        -h | --help )
+            usage
+            exit 0
+            ;;
+        * )
+            echo "Unknown argument ${1}"
+            exit 1
+            ;;
+    esac
+    shift
+done
+
+DISTRIBUTED_ARGS="
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --node_rank $NODE_RANK \
+    --master_addr $MASTER_ADDR \
+    --master_port $MASTER_PORT
+"
+
+torchrun $DISTRIBUTED_ARGS finetune.py \
+    --model_name_or_path $MODEL \
+    --data_path $DATA \
+    --bf16 True \
+    --output_dir output_qwen \
+    --num_train_epochs 5 \
+    --per_device_train_batch_size 2 \
+    --per_device_eval_batch_size 1 \
+    --gradient_accumulation_steps 8 \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 1000 \
+    --save_total_limit 10 \
+    --learning_rate 3e-4 \
+    --weight_decay 0.1 \
+    --adam_beta2 0.95 \
+    --warmup_ratio 0.01 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --report_to "none" \
+    --model_max_length 512 \
+    --lazy_preprocess True \
+    --use_lora \
+    --gradient_checkpointing \
+    --deepspeed ${DS_CONFIG_PATH}

qwen/finetune_qlora_ds.sh

@@ -0,0 +1,93 @@
+#!/bin/bash
+export CUDA_DEVICE_MAX_CONNECTIONS=1
+DIR=`pwd`
+
+# Guide:
+# This script supports distributed training on multi-gpu workers (as well as single-worker training).
+# Please set the options below according to the comments.
+# For multi-gpu workers training, these options should be manually set for each worker.
+# After setting the options, please run the script on each worker.
+
+# Number of GPUs per GPU worker
+GPUS_PER_NODE=$(python -c 'import torch; print(torch.cuda.device_count())')
+
+# Number of GPU workers, for single-worker training, please set to 1
+NNODES=${NNODES:-1}
+
+# The rank of this worker, should be in {0, ..., WORKER_CNT-1}, for single-worker training, please set to 0
+NODE_RANK=${NODE_RANK:-0}
+
+# The ip address of the rank-0 worker, for single-worker training, please set to localhost
+MASTER_ADDR=${MASTER_ADDR:localhost}
+
+# The port for communication
+MASTER_PORT=${MASTER_PORT:-6001}
+
+MODEL="Qwen/Qwen-7B-Chat-Int4" # Set the path if you do not want to load from huggingface directly
+# ATTENTION: specify the path to your training data, which should be a json file consisting of a list of conversations.
+# See the section for finetuning in README for more information.
+DATA="path_to_data"
+
+function usage() {
+    echo '
+Usage: bash finetune/finetune_qlora_ds.sh [-m MODEL_PATH] [-d DATA_PATH]
+'
+}
+
+while [[ "$1" != "" ]]; do
+    case $1 in
+        -m | --model )
+            shift
+            MODEL=$1
+            ;;
+        -d | --data )
+            shift
+            DATA=$1
+            ;;
+        -h | --help )
+            usage
+            exit 0
+            ;;
+        * )
+            echo "Unknown argument ${1}"
+            exit 1
+            ;;
+    esac
+    shift
+done
+
+DISTRIBUTED_ARGS="
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --node_rank $NODE_RANK \
+    --master_addr $MASTER_ADDR \
+    --master_port $MASTER_PORT
+"
+
+# Remember to use --fp16 instead of --bf16 due to autogptq
+torchrun $DISTRIBUTED_ARGS finetune.py \
+    --model_name_or_path $MODEL \
+    --data_path $DATA \
+    --fp16 True \
+    --output_dir output_qwen \
+    --num_train_epochs 5 \
+    --per_device_train_batch_size 2 \
+    --per_device_eval_batch_size 1 \
+    --gradient_accumulation_steps 8 \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 1000 \
+    --save_total_limit 10 \
+    --learning_rate 3e-4 \
+    --weight_decay 0.1 \
+    --adam_beta2 0.95 \
+    --warmup_ratio 0.01 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --report_to "none" \
+    --model_max_length 512 \
+    --lazy_preprocess True \
+    --use_lora \
+    --q_lora \
+    --gradient_checkpointing \
+    --deepspeed finetune/ds_config_zero2.json

qwen/finetune_qlora_single_gpu.sh

@@ -0,0 +1,66 @@
+#!/bin/bash
+export CUDA_DEVICE_MAX_CONNECTIONS=1
+DIR=`pwd`
+
+MODEL="Qwen/Qwen-7B-Chat-Int4" # Set the path if you do not want to load from huggingface directly
+# ATTENTION: specify the path to your training data, which should be a json file consisting of a list of conversations.
+# See the section for finetuning in README for more information.
+DATA="path_to_data"
+
+function usage() {
+    echo '
+Usage: bash finetune/finetune_qlora_single_gpu.sh [-m MODEL_PATH] [-d DATA_PATH]
+'
+}
+
+while [[ "$1" != "" ]]; do
+    case $1 in
+        -m | --model )
+            shift
+            MODEL=$1
+            ;;
+        -d | --data )
+            shift
+            DATA=$1
+            ;;
+        -h | --help )
+            usage
+            exit 0
+            ;;
+        * )
+            echo "Unknown argument ${1}"
+            exit 1
+            ;;
+    esac
+    shift
+done
+
+export CUDA_VISIBLE_DEVICES=0
+
+# Remember to use --fp16 instead of --bf16 due to autogptq
+python finetune.py \
+  --model_name_or_path $MODEL \
+  --data_path $DATA \
+  --fp16 True \
+  --output_dir output_qwen \
+  --num_train_epochs 5 \
+  --per_device_train_batch_size 2 \
+  --per_device_eval_batch_size 1 \
+  --gradient_accumulation_steps 8 \
+  --evaluation_strategy "no" \
+  --save_strategy "steps" \
+  --save_steps 1000 \
+  --save_total_limit 10 \
+  --learning_rate 3e-4 \
+  --weight_decay 0.1 \
+  --adam_beta2 0.95 \
+  --warmup_ratio 0.01 \
+  --lr_scheduler_type "cosine" \
+  --logging_steps 1 \
+  --report_to "none" \
+  --model_max_length 512 \
+  --lazy_preprocess True \
+  --gradient_checkpointing \
+  --use_lora \
+  --q_lora \
+  --deepspeed finetune/ds_config_zero2.json