Update README.md
Browse files
README.md
CHANGED
|
@@ -90,6 +90,17 @@ Both gpt-oss models can be fine-tuned for a variety of specialized use cases.
|
|
| 90 |
|
| 91 |
This smaller model `gpt-oss-20b` can be fine-tuned on consumer hardware, whereas the larger [`gpt-oss-120b`](https://huggingface.co/openai/gpt-oss-120b) can be fine-tuned on a single H100 node.
|
| 92 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 93 |
|
| 94 |
|
| 95 |
|
|
|
|
| 90 |
|
| 91 |
This smaller model `gpt-oss-20b` can be fine-tuned on consumer hardware, whereas the larger [`gpt-oss-120b`](https://huggingface.co/openai/gpt-oss-120b) can be fine-tuned on a single H100 node.
|
| 92 |
|
| 93 |
+
# Benchmark
|
| 94 |
+
|
| 95 |
+
| Tasks |metatune R0|metatune R1|Llama 4 Maverick|
|
| 96 |
+
|:-----------------------|:-----|:-----|:----- |
|
| 97 |
+
|gsm8k_cot |0.91 | - | - |
|
| 98 |
+
|gpqa_diamond_cot_n_shot |0.722 | | - |
|
| 99 |
+
|hellaswag |0.421 | 0.525|- |
|
| 100 |
+
|arc_challenge |0.349 | 0.349|- |
|
| 101 |
+
|winogrande |0.7851| 0.5928|- |
|
| 102 |
+
|
| 103 |
+
|
| 104 |
|
| 105 |
|
| 106 |
|