Update README.md
Browse files
README.md
CHANGED
|
@@ -7,6 +7,8 @@ The model was trained on synthetic data generated by [Glaive](https://glaive.ai)
|
|
| 7 |
## Benchmarks
|
| 8 |
|
| 9 |
|
|
|
|
|
|
|
| 10 |
Trained from Llama 3.1 70B Instruct, you can sample from Reflection 70B using the same code, pipelines, etc. as any other Llama model. It even uses the stock Llama 3.1 chat template format (though, we've trained in a few new special tokens to aid in reasoning and reflection).
|
| 11 |
|
| 12 |
During sampling, the model will start by outputting reasoning inside `<thinking>` and `</thinking>` tags, and then once it is satisfied with its reasoning, it will output the final answer inside `<output>` and `</output>` tags. Each of these tags are special tokens, trained into the model.
|
|
|
|
| 7 |
## Benchmarks
|
| 8 |
|
| 9 |
|
| 10 |
+
All benchmarks tested have been checked for contamination by running [LMSys's LLM Decontaminator](https://github.com/lm-sys/llm-decontaminator).
|
| 11 |
+
|
| 12 |
Trained from Llama 3.1 70B Instruct, you can sample from Reflection 70B using the same code, pipelines, etc. as any other Llama model. It even uses the stock Llama 3.1 chat template format (though, we've trained in a few new special tokens to aid in reasoning and reflection).
|
| 13 |
|
| 14 |
During sampling, the model will start by outputting reasoning inside `<thinking>` and `</thinking>` tags, and then once it is satisfied with its reasoning, it will output the final answer inside `<output>` and `</output>` tags. Each of these tags are special tokens, trained into the model.
|