Update README.md
Browse files
README.md
CHANGED
|
@@ -58,7 +58,7 @@ We used [OpenAI's Chat Markup Language (ChatML)](https://github.com/openai/opena
|
|
| 58 |
You are MistralOrca, a large language model trained by Alignment Lab AI. Write out your reasoning step-by-step to be sure you get the right answers!
|
| 59 |
<|im_end|>
|
| 60 |
<|im_start|>user
|
| 61 |
-
How are you
|
| 62 |
<|im_start|>assistant
|
| 63 |
I am doing well!<|im_end|>
|
| 64 |
<|im_start|>user
|
|
@@ -73,6 +73,8 @@ Please tell me about how mistral winds have attracted super-orcas.<|im_end|>
|
|
| 73 |
We have evaluated using the methodology and tools for the HuggingFace Leaderboard, and find that we have dramatically improved upon the base model.
|
| 74 |
We find **105%** of the base model's performance on HF Leaderboard evals, averaging **65.33**.
|
| 75 |
|
|
|
|
|
|
|
| 76 |
|
| 77 |
|
| 78 |
|
|
|
|
| 58 |
You are MistralOrca, a large language model trained by Alignment Lab AI. Write out your reasoning step-by-step to be sure you get the right answers!
|
| 59 |
<|im_end|>
|
| 60 |
<|im_start|>user
|
| 61 |
+
How are you?<|im_end|>
|
| 62 |
<|im_start|>assistant
|
| 63 |
I am doing well!<|im_end|>
|
| 64 |
<|im_start|>user
|
|
|
|
| 73 |
We have evaluated using the methodology and tools for the HuggingFace Leaderboard, and find that we have dramatically improved upon the base model.
|
| 74 |
We find **105%** of the base model's performance on HF Leaderboard evals, averaging **65.33**.
|
| 75 |
|
| 76 |
+
At release time, this beats all 7B models, and all but one 13B.
|
| 77 |
+
|
| 78 |
|
| 79 |
|
| 80 |
|