Model save
Browse files- README.md +77 -0
- adapter_model.safetensors +1 -1
- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +0 -0
README.md
ADDED
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: peft
|
3 |
+
tags:
|
4 |
+
- trl
|
5 |
+
- dpo
|
6 |
+
- generated_from_trainer
|
7 |
+
base_model: allenai/tulu-2-13b
|
8 |
+
model-index:
|
9 |
+
- name: tulu2-13b-cost-tulumix-1e-6
|
10 |
+
results: []
|
11 |
+
---
|
12 |
+
|
13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
14 |
+
should probably proofread and complete it, then remove this comment. -->
|
15 |
+
|
16 |
+
# tulu2-13b-cost-tulumix-1e-6
|
17 |
+
|
18 |
+
This model is a fine-tuned version of [allenai/tulu-2-13b](https://huggingface.co/allenai/tulu-2-13b) on the None dataset.
|
19 |
+
It achieves the following results on the evaluation set:
|
20 |
+
- Loss: 0.6891
|
21 |
+
- Rewards/chosen: 0.0068
|
22 |
+
- Rewards/rejected: -0.0036
|
23 |
+
- Rewards/accuracies: 0.5570
|
24 |
+
- Rewards/margins: 0.0103
|
25 |
+
- Rewards/margins Max: 0.1265
|
26 |
+
- Rewards/margins Min: -0.1089
|
27 |
+
- Rewards/margins Std: 0.0783
|
28 |
+
- Logps/rejected: -316.0208
|
29 |
+
- Logps/chosen: -324.2853
|
30 |
+
- Logits/rejected: -0.9127
|
31 |
+
- Logits/chosen: -1.0350
|
32 |
+
|
33 |
+
## Model description
|
34 |
+
|
35 |
+
More information needed
|
36 |
+
|
37 |
+
## Intended uses & limitations
|
38 |
+
|
39 |
+
More information needed
|
40 |
+
|
41 |
+
## Training and evaluation data
|
42 |
+
|
43 |
+
More information needed
|
44 |
+
|
45 |
+
## Training procedure
|
46 |
+
|
47 |
+
### Training hyperparameters
|
48 |
+
|
49 |
+
The following hyperparameters were used during training:
|
50 |
+
- learning_rate: 1e-06
|
51 |
+
- train_batch_size: 2
|
52 |
+
- eval_batch_size: 8
|
53 |
+
- seed: 42
|
54 |
+
- distributed_type: multi-GPU
|
55 |
+
- num_devices: 2
|
56 |
+
- gradient_accumulation_steps: 2
|
57 |
+
- total_train_batch_size: 8
|
58 |
+
- total_eval_batch_size: 16
|
59 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
60 |
+
- lr_scheduler_type: cosine
|
61 |
+
- lr_scheduler_warmup_ratio: 0.1
|
62 |
+
- num_epochs: 1
|
63 |
+
|
64 |
+
### Training results
|
65 |
+
|
66 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Rewards/margins Max | Rewards/margins Min | Rewards/margins Std | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
67 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:-------------------:|:-------------------:|:-------------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
68 |
+
| 0.6183 | 1.0 | 3174 | 0.6891 | 0.0068 | -0.0036 | 0.5570 | 0.0103 | 0.1265 | -0.1089 | 0.0783 | -316.0208 | -324.2853 | -0.9127 | -1.0350 |
|
69 |
+
|
70 |
+
|
71 |
+
### Framework versions
|
72 |
+
|
73 |
+
- PEFT 0.7.1
|
74 |
+
- Transformers 4.39.0.dev0
|
75 |
+
- Pytorch 2.1.2+cu121
|
76 |
+
- Datasets 2.14.6
|
77 |
+
- Tokenizers 0.15.2
|
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1001466944
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6086cf05c49855c3ace78fab213e780c6fef5241b187d543da189d66e8959c8a
|
3 |
size 1001466944
|
all_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 1.0,
|
3 |
+
"train_loss": 0.6393037238208168,
|
4 |
+
"train_runtime": 24580.6168,
|
5 |
+
"train_samples": 25396,
|
6 |
+
"train_samples_per_second": 1.033,
|
7 |
+
"train_steps_per_second": 0.129
|
8 |
+
}
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 1.0,
|
3 |
+
"train_loss": 0.6393037238208168,
|
4 |
+
"train_runtime": 24580.6168,
|
5 |
+
"train_samples": 25396,
|
6 |
+
"train_samples_per_second": 1.033,
|
7 |
+
"train_steps_per_second": 0.129
|
8 |
+
}
|
trainer_state.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|