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

.

This commit is contained in:
hailin 2025-08-28 13:12:12 +08:00
parent ea93b3c067
commit a7a54eb4dd
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
run_ds.sh
View File

@ -23,10 +23,10 @@ EVAL_DATA_GLOB="${EVAL_DATA_GLOB:-$HOME/datasets/my_corpus/test.jsonl}"
OUTDIR="${OUTDIR:-$HOME/checkpoints/run-qwen3-8b}" OUTDIR="${OUTDIR:-$HOME/checkpoints/run-qwen3-8b}"
SEQ_LEN="${SEQ_LEN:-4096}" SEQ_LEN="${SEQ_LEN:-4096}"
LR="${LR:-2e-5}" LR="${LR:-2e-5}"
GAS="${GAS:-64}" GAS="${GAS:-1}"
LOG_STEPS="${LOG_STEPS:-10}" LOG_STEPS="${LOG_STEPS:-10}"
SAVE_STEPS="${SAVE_STEPS:-500}" SAVE_STEPS="${SAVE_STEPS:-10}"
MAX_STEPS="${MAX_STEPS:-10000}" MAX_STEPS="${MAX_STEPS:-62}"
# 轻量校验（只在发起节点做；各 rank 在脚本里也会 mkdir） # 轻量校验（只在发起节点做；各 rank 在脚本里也会 mkdir）
[ -d "$MODEL_NAME_OR_PATH" ] || { echo "ERR: model not found at $MODEL_NAME_OR_PATH"; exit 1; } [ -d "$MODEL_NAME_OR_PATH" ] || { echo "ERR: model not found at $MODEL_NAME_OR_PATH"; exit 1; }