Update README.md
Browse files
README.md
CHANGED
|
@@ -36,7 +36,7 @@ This model is a LoRA-tuned variant of [gpt-oss-20b](https://huggingface.co/opena
|
|
| 36 |
It was trained on ~36M tokens from [ua-codeforces-cots-open-r1](https://huggingface.co/datasets/anon-researcher-ua/ua-codeforces-cots-open-r1) ([ua-codeforces-cots-open-r1-for-training](https://huggingface.co/datasets/anon-researcher-ua/ua-codeforces-cots-open-r1-for-training) in fact), which extends [open-r1/codeforces-cots](https://huggingface.co/datasets/open-r1/codeforces-cots) with translated Ukrainian problem statements and low/medium/high reasoning traces.
|
| 37 |
|
| 38 |
The model supports low, medium, and high reasoning effort as well as language control for reasoning via system prompts.
|
| 39 |
-
This model is able to reason in other languages as well with proper sampling params (check examples below).
|
| 40 |
|
| 41 |
# Training Details
|
| 42 |
|
|
|
|
| 36 |
It was trained on ~36M tokens from [ua-codeforces-cots-open-r1](https://huggingface.co/datasets/anon-researcher-ua/ua-codeforces-cots-open-r1) ([ua-codeforces-cots-open-r1-for-training](https://huggingface.co/datasets/anon-researcher-ua/ua-codeforces-cots-open-r1-for-training) in fact), which extends [open-r1/codeforces-cots](https://huggingface.co/datasets/open-r1/codeforces-cots) with translated Ukrainian problem statements and low/medium/high reasoning traces.
|
| 37 |
|
| 38 |
The model supports low, medium, and high reasoning effort as well as language control for reasoning via system prompts.
|
| 39 |
+
This model is able to reason in other languages as well with proper sampling params ([check examples below](https://huggingface.co/anon-researcher-ua/codegpt-oss-20b#other-languages)).
|
| 40 |
|
| 41 |
# Training Details
|
| 42 |
|