p04798526
/
LLaMA-Factory-Mirror
1
0
Fork 1
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Compare commits

base: p04798526:3981d608f52c38bbf1a8b3df3c708049999c7d70
Branches Tags
p04798526:main
p04798526:v0.8.3
p04798526:v0.8.2
p04798526:v0.8.1
p04798526:v0.8.0
p04798526:v0.7.1
p04798526:v0.7.0
p04798526:v0.6.3
p04798526:v0.6.2
p04798526:v0.6.1
p04798526:v0.6.0
p04798526:v0.5.3
p04798526:v0.5.2
p04798526:v0.5.0
p04798526:v0.4.0
p04798526:v0.3.3
p04798526:v0.3.2
p04798526:v0.3.0
p04798526:v0.2.2
p04798526:v0.2.1
p04798526:v0.2.0
p04798526:v0.1.8
p04798526:v0.1.7
p04798526:v0.1.6
p04798526:v0.1.5
p04798526:v0.1.4
p04798526:v0.1.3
p04798526:v0.1.2
p04798526:v0.1.1
p04798526:v0.1.0
p04798526:v0.0.9
...
compare: p04798526:d7d54df5253ac9116efe8da6333e585f9d3795ed
Branches Tags
p04798526:main
p04798526:v0.8.3
p04798526:v0.8.2
p04798526:v0.8.1
p04798526:v0.8.0
p04798526:v0.7.1
p04798526:v0.7.0
p04798526:v0.6.3
p04798526:v0.6.2
p04798526:v0.6.1
p04798526:v0.6.0
p04798526:v0.5.3
p04798526:v0.5.2
p04798526:v0.5.0
p04798526:v0.4.0
p04798526:v0.3.3
p04798526:v0.3.2
p04798526:v0.3.0
p04798526:v0.2.2
p04798526:v0.2.1
p04798526:v0.2.0
p04798526:v0.1.8
p04798526:v0.1.7
p04798526:v0.1.6
p04798526:v0.1.5
p04798526:v0.1.4
p04798526:v0.1.3
p04798526:v0.1.2
p04798526:v0.1.1
p04798526:v0.1.0
p04798526:v0.0.9

2 Commits
3981d608f5 ... d7d54df525

Author SHA1 Message Date
wql d7d54df525 change: change batch run 2024-08-19 10:48:31 +08:00
wql 7c5d56ca26 change: change yaml 2024-08-19 10:41:43 +08:00
2 changed files with 50 additions and 49 deletions
Show Stats Expand all files Collapse all files

96
batch_run.sh
View File

@ -1,56 +1,56 @@
FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_1.yaml > results/lora_sft/Llama2-7B/llama2_lora_sft_1.log
FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_2.yaml > results/lora_sft/Llama2-7B/llama2_lora_sft_2.log
FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_3.yaml > results/lora_sft/Llama2-7B/llama2_lora_sft_3.log
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_1_single.yaml > results/lora_sft/Llama2-7B/llama2_lora_sft_1_single.log
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_2_single.yaml > results/lora_sft/Llama2-7B/llama2_lora_sft_2_single.log
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_3_single.yaml > results/lora_sft/Llama2-7B/llama2_lora_sft_3_single.log
FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_1.yaml | tee results/lora_sft/Llama2-7B/llama2_lora_sft_1.txt
FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_2.yaml | tee results/lora_sft/Llama2-7B/llama2_lora_sft_2.txt
FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_3.yaml | tee results/lora_sft/Llama2-7B/llama2_lora_sft_3.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_1_single.yaml | tee results/lora_sft/Llama2-7B/llama2_lora_sft_1_single.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_2_single.yaml | tee results/lora_sft/Llama2-7B/llama2_lora_sft_2_single.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Llama2-7B/llama2_lora_sft_3_single.yaml | tee results/lora_sft/Llama2-7B/llama2_lora_sft_3_single.txt
FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_1.yaml > results/inference/Llama2-7B/llama2_predict_1.log
FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_2.yaml > results/inference/Llama2-7B/llama2_predict_2.log
FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_3.yaml > results/inference/Llama2-7B/llama2_predict_3.log
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_1_single.yaml > results/inference/Llama2-7B/llama2_predict_1_single.log
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_2_single.yaml > results/inference/Llama2-7B/llama2_predict_2_single.log
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_3_single.yaml > results/inference/Llama2-7B/llama2_predict_3_single.log
FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_1.yaml | tee results/inference/Llama2-7B/llama2_predict_1.txt
FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_2.yaml | tee results/inference/Llama2-7B/llama2_predict_2.txt
FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_3.yaml | tee results/inference/Llama2-7B/llama2_predict_3.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_1_single.yaml | tee results/inference/Llama2-7B/llama2_predict_1_single.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_2_single.yaml | tee results/inference/Llama2-7B/llama2_predict_2_single.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_3_single.yaml | tee results/inference/Llama2-7B/llama2_predict_3_single.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_1.yaml > results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_1.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_2.yaml > results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_2.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_3.yaml > results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_3.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_1_single.yaml > results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_1_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_2_single.yaml > results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_2_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_3_single.yaml > results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_3_single.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_1.yaml | tee results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_1.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_2.yaml | tee results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_2.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_3.yaml | tee results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_3.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_1_single.yaml | tee results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_1_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_2_single.yaml | tee results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_2_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_3_single.yaml | tee results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_3_single.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_1.yaml > results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_1.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_2.yaml > results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_2.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_3.yaml > results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_3.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_1_single.yaml > results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_1_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_2_single.yaml > results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_2_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_3_single.yaml > results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_3_single.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_1.yaml | tee results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_1.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_2.yaml | tee results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_2.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_3.yaml | tee results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_3.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_1_single.yaml | tee results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_1_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_2_single.yaml | tee results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_2_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_3_single.yaml | tee results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_3_single.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_1.yaml > results/lora_sft/Qwen-7B/Qwen_lora_sft_1.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_2.yaml > results/lora_sft/Qwen-7B/Qwen_lora_sft_2.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_3.yaml > results/lora_sft/Qwen-7B/Qwen_lora_sft_3.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_1_single.yaml > results/lora_sft/Qwen-7B/Qwen_lora_sft_1_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_2_single.yaml > results/lora_sft/Qwen-7B/Qwen_lora_sft_2_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_3_single.yaml > results/lora_sft/Qwen-7B/Qwen_lora_sft_3_single.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_1.yaml | tee results/lora_sft/Qwen-7B/Qwen_lora_sft_1.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_2.yaml | tee results/lora_sft/Qwen-7B/Qwen_lora_sft_2.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_3.yaml | tee results/lora_sft/Qwen-7B/Qwen_lora_sft_3.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_1_single.yaml | tee results/lora_sft/Qwen-7B/Qwen_lora_sft_1_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_2_single.yaml | tee results/lora_sft/Qwen-7B/Qwen_lora_sft_2_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft/Qwen-7B/Qwen_lora_sft_3_single.yaml | tee results/lora_sft/Qwen-7B/Qwen_lora_sft_3_single.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_1.yaml > results/inference/Qwen-7B/Qwen_predict_1.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_2.yaml > results/inference/Qwen-7B/Qwen_predict_2.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_3.yaml > results/inference/Qwen-7B/Qwen_predict_3.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_1_single.yaml > results/inference/Qwen-7B/Qwen_predict_1_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_2_single.yaml > results/inference/Qwen-7B/Qwen_predict_2_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_3_single.yaml > results/inference/Qwen-7B/Qwen_predict_3_single.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_1.yaml | tee results/inference/Qwen-7B/Qwen_predict_1.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_2.yaml | tee results/inference/Qwen-7B/Qwen_predict_2.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_3.yaml | tee results/inference/Qwen-7B/Qwen_predict_3.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_1_single.yaml | tee results/inference/Qwen-7B/Qwen_predict_1_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_2_single.yaml | tee results/inference/Qwen-7B/Qwen_predict_2_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_3_single.yaml | tee results/inference/Qwen-7B/Qwen_predict_3_single.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_1.yaml > results/inference/ChatGLM2-6B/ChatGLM2_predict_1.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_2.yaml > results/inference/ChatGLM2-6B/ChatGLM2_predict_2.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_3.yaml > results/inference/ChatGLM2-6B/ChatGLM2_predict_3.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_1_single.yaml > results/inference/ChatGLM2-6B/ChatGLM2_predict_1_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_2_single.yaml > results/inference/ChatGLM2-6B/ChatGLM2_predict_2_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_3_single.yaml > results/inference/ChatGLM2-6B/ChatGLM2_predict_3_single.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_1.yaml | tee results/inference/ChatGLM2-6B/ChatGLM2_predict_1.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_2.yaml | tee results/inference/ChatGLM2-6B/ChatGLM2_predict_2.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_3.yaml | tee results/inference/ChatGLM2-6B/ChatGLM2_predict_3.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_1_single.yaml | tee results/inference/ChatGLM2-6B/ChatGLM2_predict_1_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_2_single.yaml | tee results/inference/ChatGLM2-6B/ChatGLM2_predict_2_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_3_single.yaml | tee results/inference/ChatGLM2-6B/ChatGLM2_predict_3_single.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_1.yaml > results/inference/Baichuan2-7B/Baichuan2_predict_1.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_2.yaml > results/inference/Baichuan2-7B/Baichuan2_predict_2.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_3.yaml > results/inference/Baichuan2-7B/Baichuan2_predict_3.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_1_single.yaml > results/inference/Baichuan2-7B/Baichuan2_predict_1_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_2_single.yaml > results/inference/Baichuan2-7B/Baichuan2_predict_2_single.log
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_3_single.yaml > results/inference/Baichuan2-7B/Baichuan2_predict_3_single.log
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_1.yaml | tee results/inference/Baichuan2-7B/Baichuan2_predict_1.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_2.yaml | tee results/inference/Baichuan2-7B/Baichuan2_predict_2.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_3.yaml | tee results/inference/Baichuan2-7B/Baichuan2_predict_3.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_1_single.yaml | tee results/inference/Baichuan2-7B/Baichuan2_predict_1_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_2_single.yaml | tee results/inference/Baichuan2-7B/Baichuan2_predict_2_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_3_single.yaml | tee results/inference/Baichuan2-7B/Baichuan2_predict_3_single.txt

3
results/lora_sft/Llama2-7B_2/llama2_lora_sft_1_no_ms_bf.yaml → results/lora_sft/Llama2-7B_2/llama2_lora_sft_1_ms1000_bf.yaml
View File

@ -16,7 +16,7 @@ overwrite_cache: true
preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Llama2-7B/llama2_lora_sft_1
output_dir: ./results/lora_sft/Llama2-7B_2/llama2_lora_sft_1_ms1000_bf
logging_steps: 3
save_steps: 100
plot_loss: true
@ -31,6 +31,7 @@ lr_scheduler_type: cosine
warmup_ratio: 0.1
bf16: true
ddp_timeout: 180000000
max_steps: 1000
### eval
val_size: 0.1