Downloads · 30 days
13
13% of all-time downloads
pythia-160m-minipile
Marcus2112/pythia-160m-minipile
pythia-160m-minipile is a machine learning model from Marcus2112. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
All-time downloads
97
Public
Parameters
124M
495 MB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors495 MB · 99%
At a glance
- Model type
- gpt_neox
- Access
- Public
- Created
- Nov 15, 2024
- Updated
- Jan 17, 2025
- SHA
- 8c00faa9
Base models
Datasets
Model card
From the Hugging Face model README
| Benchmark | Measure | 160M MiniPile | 160M Pile Deduplicated | |
|---|---|---|---|---|
| ARC-Challenge | acc | ↑ | 0.2125 ± 0.0120 | 0.1997 ± 0.0117 |
| MMLU | acc | ↑ | 0.2699 ± 0.0037 | 0.2299 ± 0.0035 |
| HellaSwag | acc | ↑ | 0.2560 ± 0.0044 | 0.2903 ± 0.0045 |
| WinoGrande | acc | ↑ | 0.4720 ± 0.0140 | 0.4964 ± 0.0141 |
| Lambada (OpenAI) | acc | ↑ | 0.0000 ± 0.0000 | 0.3689 ± 0.0067 |
| Lambada (OpenAI) | perplexity | ↓ | 3033175.2693 ± 288926.5827 | 31.2589 ± 1.1594 |
| Lambada (Std) | acc | ↑ | 0.0000 ± 0.0000 | 0.2335 ± 0.0059 |
| Lambada (Std) | perplexity | ↓ | 27067951.3461 ± 2710040.1910 | 172.7619 ± 7.7265 |
| BLiMP | acc | ↑ | 0.5194 ± 0.0018 | 0.7294 ± 0.0015 |
- Type
- gpt_neox
- Languages
- en
- Created
- Nov 15, 2024
- Updated
- Jan 17, 2025