theblackcat102
commited on
Commit
•
60acef3
1
Parent(s):
b5ecb22
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,38 @@
|
|
1 |
---
|
2 |
license: mit
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: mit
|
3 |
+
datasets:
|
4 |
+
- OpenAssistant/oasst1
|
5 |
---
|
6 |
+
|
7 |
+
[wandb](https://wandb.ai/open-assistant/reward-model/runs/rwq6bnhw?workspace=)
|
8 |
+
|
9 |
+
```
|
10 |
+
test_rm_labeling:
|
11 |
+
is_reward_model: true
|
12 |
+
pooling: last
|
13 |
+
sort_by_length: false
|
14 |
+
use_custom_sampler: true
|
15 |
+
model_name: microsoft/deberta-v3-large
|
16 |
+
learning_rate: 3e-5
|
17 |
+
residual_dropout: 0.0
|
18 |
+
weight_decay: 0.0
|
19 |
+
max_length: 2048
|
20 |
+
use_flash_attention: true
|
21 |
+
gradient_checkpointing: true
|
22 |
+
warmup_steps: 50
|
23 |
+
dtype: float16
|
24 |
+
gradient_accumulation_steps: 20
|
25 |
+
per_device_train_batch_size: 1
|
26 |
+
per_device_eval_batch_size: 4
|
27 |
+
num_train_epochs: 3
|
28 |
+
eval_steps: 251
|
29 |
+
save_steps: 500
|
30 |
+
loss_fn: HybridRMLoss
|
31 |
+
datasets:
|
32 |
+
- oasst_export_w_label:
|
33 |
+
lang: "bg,ca,cs,da,de,en,es,fr,hr,hu,it,nl,pl,pt,ro,ru,sl,sr,sv,uk,zh,ja,th,vi"
|
34 |
+
input_file_path: 2023-04-12_oasst_release_ready_synth.jsonl.gz
|
35 |
+
input_label_path: 2023-04-12_oasst_all.messages.jsonl.gz
|
36 |
+
val_split: 0.1
|
37 |
+
```
|
38 |
+
|