Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
datasets:
|
| 4 |
+
- allenai/dolma
|
| 5 |
+
- allenai/tulu-v2-sft-mixture-olmo-4096
|
| 6 |
+
- argilla/ultrafeedback-binarized-preferences-cleaned
|
| 7 |
+
language:
|
| 8 |
+
- en
|
| 9 |
+
---
|
| 10 |
+
# OLMo-1B-0724 Instruct
|
| 11 |
+
|
| 12 |
+
This is a version of [OLMo-1B-0724-hf](https://huggingface.co/allenai/OLMo-1B-0724-hf) that has undergone SFT and DPO training.
|
| 13 |
+
See [the SFT model card for details on SFT training](https://huggingface.co/hamishivi/OLMo-1B-0724-SFT-hf).
|
| 14 |
+
|
| 15 |
+
This model is initialised from [OLMo-1B-0724-SFT-hf](https://huggingface.co/hamishivi/OLMo-1B-0724-SFT-hf), and then DPO trained on a cleaned ultrafeedback dataset
|
| 16 |
+
for 3 epochs with a batch size of 32, beta of 0.1, linear warmup for 10% of training, and then linear cooldown.
|
| 17 |
+
|
| 18 |
+
Evals are as follows:
|
| 19 |
+
|
| 20 |
+
| Metric | [OLMo-1B-0724-hf](https://huggingface.co/allenai/OLMo-1B-0724-hf) | [OLMo-1B-0724-SFT-hf](https://huggingface.co/hamishivi/OLMo-1B-0724-SFT-hf) | **[OLMo-1B-0724-Instruct-hf](https://huggingface.co/hamishivi/OLMo-1B-0724-Instruct-hf) (this model!)** |
|
| 21 |
+
|---------------------------|-----------------|---------------------|-------------------------|
|
| 22 |
+
| MMLU 0-shot | 25.0 | 36.0 | **36.7** |
|
| 23 |
+
| GSM8k CoT 8-shot | 7.0 | **12.5** | **12.5** |
|
| 24 |
+
| BBH CoT 3-shot | 22.5 | 27.2 | **30.6** |
|
| 25 |
+
| HumanEval P@10 | 16.0 | 21.2 | **22.0** |
|
| 26 |
+
| AlpacaEval 1 | - | 41.5 | **50.9** |
|
| 27 |
+
| AlpacaEval 2 LC | - | **2.7** | 2.5 |
|
| 28 |
+
| Toxigen % Toxic | 80.3 | 59.7 | **14.1** |
|
| 29 |
+
| TruthfulQA %Info+True | 23.0 | 40.9 | **42.2** |
|
| 30 |
+
| IFEval Loose Acc | 20.5 | **26.1** | 24.2 |
|
| 31 |
+
| XSTest F1 | 67.6 | **81.9** | 79.8 |
|
| 32 |
+
| **Average of above metrics** | 25.2 | 33.0 | **38.7** |
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
Model training and evaluation was performed using [Open-instruct](https://github.com/allenai/open-instruct), so check that out for more details on evaluation.
|