Not useable but updating the model after each checkpoint.It is only being trained on Tinystories at the moment. This is all done on CPU only for pretraining. At checkpoint 120 not even at 0.05 of an epoch the eval harness scores are
| Tasks | Version | Filter | n-shot | Metric | Value | Stderr | ||
|---|---|---|---|---|---|---|---|---|
| arc_challenge | 1 | none | 0 | acc | ↑ | 0.1869 | ± | 0.0114 |
| none | 0 | acc_norm | ↑ | 0.2500 | ± | 0.0127 | ||
| arc_easy | 1 | none | 0 | acc | ↑ | 0.2479 | ± | 0.0089 |
| none | 0 | acc_norm | ↑ | 0.2483 | ± | 0.0089 | ||
| boolq | 2 | none | 0 | acc | ↑ | 0.3783 | ± | 0.0085 |
| hellaswag | 1 | none | 0 | acc | ↑ | 0.2501 | ± | 0.0043 |
| none | 0 | acc_norm | ↑ | 0.2440 | ± | 0.0043 | ||
| openbookqa | 1 | none | 0 | acc | ↑ | 0.1220 | ± | 0.0147 |
| none | 0 | acc_norm | ↑ | 0.1800 | ± | 0.0172 | ||
| piqa | 1 | none | 0 | acc | ↑ | 0.5114 | ± | 0.0117 |
| none | 0 | acc_norm | ↑ | 0.4820 | ± | 0.0117 | ||
| winogrande | 1 | none | 0 | acc | ↑ | 0.4925 | ± | 0.0141 |
- Downloads last month
- 377
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.