File size: 2,237 Bytes
72f4b89
 
daea860
72f4b89
 
 
 
 
 
 
 
daea860
 
 
 
72f4b89
 
daea860
 
 
72f4b89
 
daea860
e457c3e
af1eff4
daea860
d96f794
 
 
 
 
 
 
 
 
72f4b89
 
d1f3012
72f4b89
 
 
daea860
72f4b89
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
daea860
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
---
library_name: transformers
license: apache-2.0
base_model: Qwen/Qwen2.5-32B-Instruct
tags:
- llama-factory
- full
- generated_from_trainer
model-index:
- name: original
  results: []
language:
- en
datasets:
- bespokelabs/Bespoke-Stratos-17k
---

<p align="center">
    <img src="https://huggingface.co./bespokelabs/Bespoke-MiniCheck-7B/resolve/main/Bespoke-Labs-Logo.png" width="550">
</p>

## Model description
This model is a fine-tuned version of [Qwen/Qwen2.5-32B-Instruct](https://huggingface.co./Qwen/Qwen2.5-32B-Instruct) on the [Bespoke-Stratos-17k dataset](https://huggingface.co./datasets/bespokelabs/Bespoke-Stratos-17k).
The dataset is derived by distilling DeepSeek-R1 using the data pipeline of Berkeley NovaSky’s Sky-T1 with some modifications. More info in the dataset card at [Bespoke-Stratos-17k](https://huggingface.co./datasets/bespokelabs/Bespoke-Stratos-17k).
It outperforms Qwen-2.5-32B-Instruct on reasoning benchmarks:

| Metric | Bespoke-Stratos-32B | Sky-T1-32B | o1-preview | DeepSeek-R1 | DeepSeek-R1-Distill-Qwen-32B (Ours // Reported)|
|---|---|---|---|---|---|
| AIME2024 | 63.3 | 43.3 | 40.0 | 79.8 | 66.7 // 72.6 |
| MATH500 | 93.0 | 82.4 | 81.4 | 97.3 | 89.8 // 94.3 |
| GPQA-Diamond | 58.1 | 56.8 | 75.2 | 71.5 | 61.1 // 62.1 |
| LCB v2 Easy | 96.7 | 86.3 | 92.9 | - | 91.2 // - |
| LCB v2 Medium | 75.2 | 56.8 | 54.9 | - | 75.7 // - |
| LCB v2 Hard | 26.2 | 17.9 | 16.3 | - | 38.2 // - |
| LCB v2 All | 71.1 | 57.9 | 59.1 | - | 72.2 // - |

## Intended uses & limitations
Apache 2.0 License


## Training procedure
We used 8xH100 to train the model for 27 hours.

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 1
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 12
- total_train_batch_size: 96
- total_eval_batch_size: 64
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3.0

### Training results



### Framework versions

- Transformers 4.46.1
- Pytorch 2.5.1+cu124
- Datasets 3.1.0
- Tokenizers 0.20.3