Hugging Face's logo

perplexity-ai
/
r1-1776-distill-llama-70b

Text Generation
Transformers
Safetensors
llama
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
8
eugenhotaj-ppl commited on
Commit
43dd5f9
·
verified ·
1 Parent(s): 07ce166

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +7 -5
README.md CHANGED
@@ -24,12 +24,14 @@ We also ensured that the model’s math and reasoning abilities remained intact
24
  Evaluations on multiple benchmarks showed that our post-trained model performed on par with the base R1 model,
25
  indicating that the decensoring had no impact on its core reasoning capabilities.
26
 
27
- | Benchmark | R1-Distill-Llama3-70B | R1-1776-Distill-Llama3-70B |
28
  | --- | --- | --- |
29
  | China Censorship | 80.53 | 0.2 |
30
- | Internal Benchmarks (avg) | 48.4 | 47.64 |
31
  | AIME 2024 | 70 | 70 |
32
  | MATH-500 | 94.5 | 94.8 |
33
- | MMLU | 88.52 | 88.20 |
34
- | DROP | 84.55 | 84.83 |
35
- | GPQA | 65.2 | 65.05 |
 
 
 
24
  Evaluations on multiple benchmarks showed that our post-trained model performed on par with the base R1 model,
25
  indicating that the decensoring had no impact on its core reasoning capabilities.
26
 
27
+ | Benchmark | R1-Distill-Llama-70B | R1-1776-Distill-Llama-70B |
28
  | --- | --- | --- |
29
  | China Censorship | 80.53 | 0.2 |
30
+ | Internal Benchmarks (avg) | 47.64 | 48.4 |
31
  | AIME 2024 | 70 | 70 |
32
  | MATH-500 | 94.5 | 94.8 |
33
+ | MMLU | 88.52 * | 88.20 |
34
+ | DROP | 84.55 * | 84.83 |
35
+ | GPQA | 65.2 | 65.05 |
36
+
37
+ \* Evaluated by Perplexity AI since they were not reported in the [paper](https://arxiv.org/abs/2501.12948).