hailin
/
jd_train
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

.

This commit is contained in:
hailin 2025-09-09 09:24:23 +08:00
parent ff918dc8ed
commit f0c606c5d1
3 changed files with 3 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
.deepspeed_env
View File

@ -2,6 +2,6 @@ WANDB_BASE_URL=https://wandb.szaiai.com
WANDB_API_KEY=local-701636f51b4741d3862007df5cf7f12cca53d8d1 WANDB_API_KEY=local-701636f51b4741d3862007df5cf7f12cca53d8d1
WANDB_PROJECT=ds-qwen3 WANDB_PROJECT=ds-qwen3
WANDB_GROUP=q3-1.7b-ds4-2025-09-05 WANDB_GROUP=q3-1.7b-ds4-2025-09-05
WANDB_RUN_ID=q3-31.7b-lr2e-5-train1 WANDB_RUN_ID=q3-31.7b-lr2e-5-train2
WANDB_RESUME=allow WANDB_RESUME=allow
export WANDB_DIR=/tmp/$USER/wandb export WANDB_DIR=/tmp/$USER/wandb

4
mm-zero3.sh
View File

@ -31,11 +31,11 @@ deepspeed --hostfile hostfile \
--model_name_or_path /home/test/Qwen3-32B \ --model_name_or_path /home/test/Qwen3-32B \
--data_glob "/home/test/datasets/my_corpus/train.jsonl" \ --data_glob "/home/test/datasets/my_corpus/train.jsonl" \
--output_dir /home/test/checkpoints/q3-32b-ds4 \ --output_dir /home/test/checkpoints/q3-32b-ds4 \
--seq_len 1024 \ --seq_len 512 \
--per_device_train_batch_size 1 \ --per_device_train_batch_size 1 \
--gradient_accumulation_steps 1 \ --gradient_accumulation_steps 1 \
--learning_rate 2e-5 --weight_decay 0.1 --warmup_ratio 0.02 \ --learning_rate 2e-5 --weight_decay 0.1 --warmup_ratio 0.02 \
--max_steps 62 \ --max_steps 20 \
--log_interval 1 \ --log_interval 1 \
--gradient_checkpointing \ --gradient_checkpointing \
--bf16 \ --bf16 \

3
train_sft_lora.py
View File

@ -1,6 +1,3 @@
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
import os import os
os.environ.pop("PYTHONNOUSERSITE", None) os.environ.pop("PYTHONNOUSERSITE", None)
os.environ.setdefault("TOKENIZERS_PARALLELISM", "false") os.environ.setdefault("TOKENIZERS_PARALLELISM", "false")