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

change: change yaml

This commit is contained in:
wql 2024-08-22 11:21:19 +08:00
parent f47d38717f
commit 8eb67cb9f2
25 changed files with 96 additions and 72 deletions
Show Stats Download Patch File Download Diff File 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_2/Llama2-7B/llama2_lora_sft_1.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_1.txt
FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Llama2-7B/llama2_lora_sft_2.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_2.txt
FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Llama2-7B/llama2_lora_sft_3.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_3.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Llama2-7B/llama2_lora_sft_1_single.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_1_single.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Llama2-7B/llama2_lora_sft_2_single.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_2_single.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Llama2-7B/llama2_lora_sft_3_single.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_3_single.txt
FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Llama2-7B/llama2_lora_sft_1.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_1_log.txt
FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Llama2-7B/llama2_lora_sft_2.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_2_log.txt
FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Llama2-7B/llama2_lora_sft_3.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_3_log.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Llama2-7B/llama2_lora_sft_1_single.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_1_single_log.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Llama2-7B/llama2_lora_sft_2_single.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_2_single_log.txt
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Llama2-7B/llama2_lora_sft_3_single.yaml | tee results/lora_sft_2/Llama2-7B/llama2_lora_sft_3_single_log.txt
# 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/inference/Llama2-7B/llama2_predict_1.yaml | tee results/inference/Llama2-7B/llama2_predict_1_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_2.yaml | tee results/inference/Llama2-7B/llama2_predict_2_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Llama2-7B/llama2_predict_3.yaml | tee results/inference/Llama2-7B/llama2_predict_3_log.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_log.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_log.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_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_1.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_1.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_2.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_2.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_3.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_3.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_1_single.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_1_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_2_single.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_2_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_3_single.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_3_single.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_1.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_1_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_2.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_2_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_3.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_3_log.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_1_single.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_1_single_log.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_2_single.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_2_single_log.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_3_single.yaml | tee results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_3_single_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_1.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_1.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_2.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_2.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_3.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_3.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_1_single.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_1_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_2_single.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_2_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_3_single.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_3_single.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_1.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_1_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_2.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_2_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_3.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_3_log.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_1_single.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_1_single_log.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_2_single.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_2_single_log.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_3_single.yaml | tee results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_3_single_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_1.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_1.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_2.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_2.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_3.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_3.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_1_single.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_1_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_2_single.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_2_single.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_3_single.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_3_single.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_1.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_1_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_2.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_2_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_3.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_3_log.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_1_single.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_1_single_log.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_2_single.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_2_single_log.txt
# CUDA_VISIBLE_DEVICES=0 llamafactory-cli train results/lora_sft_2/Qwen-7B/Qwen_lora_sft_3_single.yaml | tee results/lora_sft_2/Qwen-7B/Qwen_lora_sft_3_single_log.txt
# 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/Qwen-7B/Qwen_predict_1.yaml | tee results/inference/Qwen-7B/Qwen_predict_1_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_2.yaml | tee results/inference/Qwen-7B/Qwen_predict_2_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Qwen-7B/Qwen_predict_3.yaml | tee results/inference/Qwen-7B/Qwen_predict_3_log.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_log.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_log.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_log.txt
# 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/ChatGLM2-6B/ChatGLM2_predict_1.yaml | tee results/inference/ChatGLM2-6B/ChatGLM2_predict_1_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_2.yaml | tee results/inference/ChatGLM2-6B/ChatGLM2_predict_2_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/ChatGLM2-6B/ChatGLM2_predict_3.yaml | tee results/inference/ChatGLM2-6B/ChatGLM2_predict_3_log.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_log.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_log.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_log.txt
# 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
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_1.yaml | tee results/inference/Baichuan2-7B/Baichuan2_predict_1_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_2.yaml | tee results/inference/Baichuan2-7B/Baichuan2_predict_2_log.txt
# FORCE_TORCHRUN=1 llamafactory-cli train results/inference/Baichuan2-7B/Baichuan2_predict_3.yaml | tee results/inference/Baichuan2-7B/Baichuan2_predict_3_log.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_log.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_log.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_log.txt

3
results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_1.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_1
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_1_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_1_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_2.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_2
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_2_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_2_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_3.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_3
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Baichuan2-7B/Baichuan2_lora_sft_3_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Baichuan2-7B/Baichuan2_lora_sft_3_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_1.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_1
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_1_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_1_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_2.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_2
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_2_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_2_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_3.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_3
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/ChatGLM2-6B/ChatGLM2_lora_sft_3_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/ChatGLM2-6B/ChatGLM2_lora_sft_3_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Llama2-7B/llama2_lora_sft_1.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Llama2-7B/llama2_lora_sft_1
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ fp16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Llama2-7B/llama2_lora_sft_1_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Llama2-7B/llama2_lora_sft_1_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ fp16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Llama2-7B/llama2_lora_sft_2.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Llama2-7B/llama2_lora_sft_2
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ fp16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Llama2-7B/llama2_lora_sft_2_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Llama2-7B/llama2_lora_sft_2_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ fp16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Llama2-7B/llama2_lora_sft_3.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Llama2-7B/llama2_lora_sft_3
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ fp16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Llama2-7B/llama2_lora_sft_3_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Llama2-7B/llama2_lora_sft_3_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ fp16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Qwen-7B/Qwen_lora_sft_1.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Qwen-7B/Qwen_lora_sft_1
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Qwen-7B/Qwen_lora_sft_1_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Qwen-7B/Qwen_lora_sft_1_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Qwen-7B/Qwen_lora_sft_2.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Qwen-7B/Qwen_lora_sft_2
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Qwen-7B/Qwen_lora_sft_2_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Qwen-7B/Qwen_lora_sft_2_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Qwen-7B/Qwen_lora_sft_3.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Qwen-7B/Qwen_lora_sft_3
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1

3
results/lora_sft_2/Qwen-7B/Qwen_lora_sft_3_single.yaml
View File

@ -18,7 +18,7 @@ preprocessing_num_workers: 16
### output
output_dir: ./results/lora_sft/Qwen-7B/Qwen_lora_sft_3_single
logging_steps: 3
save_steps: 100
save_steps: 500
plot_loss: true
overwrite_output_dir: true
@ -33,6 +33,7 @@ bf16: true
ddp_timeout: 180000000
max_steps: 1000
include_num_input_tokens_seen: true
include_tokens_per_second: true
### eval
val_size: 0.1