Philipp-Sc commited on
Commit
96aa9b7
1 Parent(s): 209c10c

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +5 -3
README.md CHANGED
@@ -15,15 +15,14 @@ base_model: mistralai/Mistral-7B-v0.2
15
  (>=checkpoint-v2)
16
  ---
17
 
18
- ## Reverse Instruct LoRa Adapter
19
 
20
- This LoRa Adapter is fine tuned to reverse engineer the original prompt of a given LLM output/response.
21
 
22
  ## Response Format
23
 
24
  "[INST]\n### System:\n{system}\n### Instruction:\n{instruction}\n[/INST]\n"
25
 
26
- (without the "")
27
 
28
  ## Prompt Template
29
 
@@ -48,6 +47,9 @@ CUDA_VISIBLE_DEVICES=0 WANDB_DISABLED=True python LLaMA-Factory/src/train_bash.p
48
  --stage sft \
49
  --model_name_or_path model_name_or_path \
50
  --checkpoint_dir checkpoint_dir \
 
 
 
51
  --do_train \
52
  --dataset default \
53
  --template vanilla \
 
15
  (>=checkpoint-v2)
16
  ---
17
 
18
+ ## Reverse Instruct LoRA Adapter
19
 
20
+ This LoRA Adapter is fine tuned to reverse engineer the original prompt of a given LLM output/response.
21
 
22
  ## Response Format
23
 
24
  "[INST]\n### System:\n{system}\n### Instruction:\n{instruction}\n[/INST]\n"
25
 
 
26
 
27
  ## Prompt Template
28
 
 
47
  --stage sft \
48
  --model_name_or_path model_name_or_path \
49
  --checkpoint_dir checkpoint_dir \
50
+ --flash_attn \
51
+ --shift_attn \
52
+ --neftune_noise_alpha 5 \
53
  --do_train \
54
  --dataset default \
55
  --template vanilla \