Update README.md
Browse files
README.md
CHANGED
@@ -45,14 +45,14 @@ As of 27 Sept 2024, Llama-3.1-Nemotron-70B-Reward performs best Overall on Rewar
|
|
45 |
| Meta-Llama-3.1-70B-Instruct | Not fully disclosed | 84.0 | 97.2 | 70.2 | 82.8 | 86.0 |
|
46 |
|
47 |
|
48 |
-
To better understand why
|
49 |
-
On the other hand, when GPT-4 annotations are used as Ground-Truth,
|
50 |
|
51 |
|
52 |
|
53 |
| Model | Type of Data Used For Training | Chat-Hard | LLMBar-Adversarial-Manual | LLMBar-Adversarial-Neighbour | LLMBar-Natural | LLMBar-Adversarial-GPTInst | LLMBar-Adversarial-GPTOut | MT-Bench-Hard|
|
54 |
|:-----------------------------|:----------------|:-----|:----------|:-------|:----------|:-----------------------|:-----------------------|:-----------------------|
|
55 |
-
|||| Human as Ground Truth | Human as Ground Truth | Human as Ground Truth |
|
56 |
| Llama-3.1-Nemotron-70B-Reward |Permissive Licensed Data Only (CC-BY-4.0) | 85.8 | 76.1 | 88.8 | 95.0 | 87.0 | 72.3 | 75.7
|
57 |
| Skywork-Reward-Gemma-2-27B | Includes GPT4 Generated Data | 91.4 | 78.3 | 89.6 | 96.0 | 97.8 | 91.5 | 86.5|
|
58 |
|
|
|
45 |
| Meta-Llama-3.1-70B-Instruct | Not fully disclosed | 84.0 | 97.2 | 70.2 | 82.8 | 86.0 |
|
46 |
|
47 |
|
48 |
+
To better understand why Llama-3.1-Nemotron-70B-Reward does less well in the Chat-Hard category, we analyze the scores for each consistutent subset under the Chat-Hard category. We find that on categories that uses human annotations as ground truth, Llama-3.1-Nemotron-70B-Reward performs similar to Skywork-Reward-Gemma-2-27B (<= 2.2% difference).
|
49 |
+
On the other hand, when GPT-4 annotations are used as Ground-Truth, Llama-3.1-Nemotron-70B-Reward trails substantially behind Skywork-Reward-Gemma-2-27B (10.8 to 19.2%). This suggests that Skywork-Reward-Gemma-2-27B can better modelling GPT-4 preference (than human-annotated preferences), likely contributed by the inclusion of GPT-4 annotated training data used to train it found in the [OffSetBias dataset](https://huggingface.co/datasets/NCSOFT/offsetbias) as part of the [Skywork-Reward-Preference-80k](https://huggingface.co/datasets/Skywork/Skywork-Reward-Preference-80K-v0.1).
|
50 |
|
51 |
|
52 |
|
53 |
| Model | Type of Data Used For Training | Chat-Hard | LLMBar-Adversarial-Manual | LLMBar-Adversarial-Neighbour | LLMBar-Natural | LLMBar-Adversarial-GPTInst | LLMBar-Adversarial-GPTOut | MT-Bench-Hard|
|
54 |
|:-----------------------------|:----------------|:-----|:----------|:-------|:----------|:-----------------------|:-----------------------|:-----------------------|
|
55 |
+
|||| Human as Ground Truth | Human as Ground Truth | Human as Ground Truth | _GPT-4 as Ground Truth_ |_GPT-4 as Ground Truth_ | _GPT-4 as Ground Truth_ |
|
56 |
| Llama-3.1-Nemotron-70B-Reward |Permissive Licensed Data Only (CC-BY-4.0) | 85.8 | 76.1 | 88.8 | 95.0 | 87.0 | 72.3 | 75.7
|
57 |
| Skywork-Reward-Gemma-2-27B | Includes GPT4 Generated Data | 91.4 | 78.3 | 89.6 | 96.0 | 97.8 | 91.5 | 86.5|
|
58 |
|