Update README.md
Browse files
README.md
CHANGED
|
@@ -34,6 +34,11 @@ output = generator([{"role": "user", "content": question}], max_new_tokens=128,
|
|
| 34 |
print(output["generated_text"])
|
| 35 |
```
|
| 36 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
## Training procedure
|
| 38 |
|
| 39 |
SFT dataset mixture:
|
|
@@ -61,8 +66,9 @@ warmup_ratio: 0.1
|
|
| 61 |
|
| 62 |
Hardware: 2 x H100 GPUs.
|
| 63 |
|
| 64 |
-
|
| 65 |
-
|
|
|
|
| 66 |
|
| 67 |
### Framework versions
|
| 68 |
|
|
@@ -71,3 +77,13 @@ Hardware: 2 x H100 GPUs.
|
|
| 71 |
- Pytorch: 2.4.1
|
| 72 |
- Datasets: 3.1.0
|
| 73 |
- Tokenizers: 0.20.3
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
print(output["generated_text"])
|
| 35 |
```
|
| 36 |
|
| 37 |
+
## Evals
|
| 38 |
+
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
|
| 42 |
## Training procedure
|
| 43 |
|
| 44 |
SFT dataset mixture:
|
|
|
|
| 66 |
|
| 67 |
Hardware: 2 x H100 GPUs.
|
| 68 |
|
| 69 |
+
_This work was performed on the HoreKa supercomputer funded by the
|
| 70 |
+
Ministry of Science, Research and the Arts Baden-Württemberg and by
|
| 71 |
+
the Federal Ministry of Education and Research._
|
| 72 |
|
| 73 |
### Framework versions
|
| 74 |
|
|
|
|
| 77 |
- Pytorch: 2.4.1
|
| 78 |
- Datasets: 3.1.0
|
| 79 |
- Tokenizers: 0.20.3
|
| 80 |
+
|
| 81 |
+
## Credits
|
| 82 |
+
|
| 83 |
+
This work wouldn't be possible without all the **great contributions from the open LLM community**. Thank you! Special kudos go to
|
| 84 |
+
|
| 85 |
+
- @philschmid for his latest [fine-tuning boilerplate](https://www.philschmid.de/fine-tune-llms-in-2025)
|
| 86 |
+
- @lvwerra, @lewtun et al for building and maintaining [trl](https://github.com/huggingface/trl)
|
| 87 |
+
- @cognitivecomputations for sharing [spectrum](https://github.com/cognitivecomputations/spectrum/tree/main)
|
| 88 |
+
|
| 89 |
+
|