ssaraf1 commited on
Commit
a0ce2e0
·
verified ·
1 Parent(s): 5412d6a

Upload adapter_config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. adapter_config.json +34 -0
adapter_config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "base_model": "Qwen/Qwen2.5-7B-Instruct",
3
+ "lora_parameters": {
4
+ "rank": 16,
5
+ "scale": 2.0,
6
+ "dropout": 0.02
7
+ },
8
+ "num_layers": 28,
9
+ "fine_tune_type": "lora",
10
+ "training": {
11
+ "stage_a": {
12
+ "iters": 800,
13
+ "data": "554K multi-workflow policy-corrected"
14
+ },
15
+ "stage_b": {
16
+ "iters": 100,
17
+ "data": "20K contrastive alignment"
18
+ },
19
+ "learning_rate": 3e-05,
20
+ "batch_size": 4,
21
+ "max_seq_length": 512,
22
+ "mask_prompt": true,
23
+ "seed": 42
24
+ },
25
+ "performance": {
26
+ "total": "42/76 (55.3%)",
27
+ "next": "10/22 (45%)",
28
+ "retry": "7/12 (58%)",
29
+ "fork": "13/14 (93%)",
30
+ "join": "10/15 (67%)",
31
+ "meta": "2/13 (15%)",
32
+ "beats_gpt41": true
33
+ }
34
+ }