| Commit message (Expand) | Author | Age | Files | Lines | |
|---|---|---|---|---|---|
| * | Use platform specific defaults for CUDA device ID | Jonas Knobloch | 2026-04-23 | 1 | -1/+13 |
| * | Destroy session on model teardown | Jonas Knobloch | 2026-04-10 | 1 | -11/+13 |
| * | Refactor model options | Jonas Knobloch | 2026-04-08 | 1 | -11/+7 |
| * | Refactor model config | Jonas Knobloch | 2026-04-08 | 1 | -6/+6 |
| * | Batch scoring | Jonas Knobloch | 2026-04-07 | 1 | -2/+4 |
| * | Rename log probs output | Jonas Knobloch | 2026-04-07 | 1 | -2/+2 |
| * | Handle environment initialization on package level | Jonas Knobloch | 2026-04-06 | 1 | -9/+1 |
| * | Require cache when generating new tokens | Jonas Knobloch | 2026-04-06 | 1 | -0/+4 |
| * | Add parameter to control logits output binding | Jonas Knobloch | 2026-04-06 | 1 | -5/+7 |
| * | Access outputs by name | Jonas Knobloch | 2026-04-06 | 1 | -9/+3 |
| * | Bind either logits or log probs | Jonas Knobloch | 2026-04-04 | 1 | -1/+1 |
| * | Add score method | Jonas Knobloch | 2026-04-04 | 1 | -10/+42 |
| * | Add config struct | Jonas Knobloch | 2026-04-04 | 1 | -14/+8 |
| * | Add allocator abstraction | Jonas Knobloch | 2026-04-04 | 1 | -177/+32 |
| * | Remove llm import | Jonas Knobloch | 2026-03-17 | 1 | -2/+0 |
| * | Use vanity import path | Jonas Knobloch | 2026-03-16 | 1 | -2/+2 |
| * | Update module paths | Jonas Knobloch | 2026-02-28 | 1 | -1/+2 |
| * | Verify sequence length | Jonas Knobloch | 2026-02-16 | 1 | -4/+9 |
| * | Allow controlling IntraOpNumThreads | Jonas Knobloch | 2026-02-11 | 1 | -6/+33 |
| * | Use seen tokens to warm up cache | Jonas Knobloch | 2025-11-26 | 1 | -47/+77 |
| * | Use dynamic session | Jonas Knobloch | 2025-11-24 | 1 | -62/+123 |
| * | Add perplexity evaluation | Jonas Knobloch | 2025-11-18 | 1 | -13/+26 |
| * | Scaffold byte-pair correction | Jonas Knobloch | 2025-11-17 | 1 | -0/+264 |
