What's Changed
- delete redundant code by @maxjeblick in #271
- Token IDs (again) by @psinger in #265
- end user shall use "make llmstudio" to run without wave checking for file changes by @pascal-pfeiffer in #269
- Fix neptune version by @psinger in #273
- Minor model template changes by @psinger in #274
- Minor doc improvements by @sherenem in #261
- dont compute np.mean(val_losses) twice by @Quetzalcohuatl in #275
- account correctly for day of year by @pascal-pfeiffer in #277
- Package upgrades and minor changes by @psinger in #291
- Removal of llama force slow by @psinger in #292
- Add data sanity checks by @maxjeblick in #289
- LLama2 config fix by @psinger in #294
- Set huggingface token when loading tokenizers by @maxjeblick in #296
- Use unk token for pad if available by @psinger in #298
- Refactor problem type by @maxjeblick in #293
- Add required URLs to LLM Studio installation guide by @sherenem in #310
- Show git version in error tab by @maxjeblick in #306
- Fix Runpod support / fix copy_config to use output dir config (#280) by @Glavin001 in #281
New Contributors
- @Quetzalcohuatl made their first contribution in #275
Full Changelog: v0.0.5...v0.0.6