v1.5.16

What's Changed

• Hotfix: remove DPO hyperparameter validations by @sbassam in #329

Full Changelog: v1.5.15...v1.5.16

v1.5.15

What's Changed

• Fix simpo_gamma default value for CLI by @timofeev1995 in #328

Full Changelog: v1.5.14...v1.5.15

v1.5.14

What's Changed

• New options for preference tuning: rpo alpha, logprobs normalization, reference-free, simpo gamma by @timofeev1995 in #327

Full Changelog: v1.5.13...v1.5.14

v1.5.13

What's Changed

• Update default LR scheduler to cosine with 0.5 cycles by @sbassam in #322
• Add support for Batch API by @rajasbansal in #325
• bump version to 1.5.13 by @orangetin in #326

New Contributors

• @sbassam made their first contribution in #322
• @rajasbansal made their first contribution in #325

Full Changelog: v1.5.11...v1.5.13

v1.5.11

What's Changed

• Accept strings for checkpoint type on download by @nbroad1881 in #308
• Migrate train_on_inputs to sft-specific params by @connermanuel in #297
• Make it possible to exclude pyarrow dep by @scosman in #276
• fix failing tests: replace mixtral test model with scout by @orangetin in #319
• Add validation for lora_dropout by @timofeev1995 in #316

New Contributors

• @scosman made their first contribution in #276

Full Changelog: 1.5.8...v1.5.11

v1.5.8

What's Changed

• add enum for tool choice required by @nbroad1881 in #309
• Use /checkpoints instead of events parsing by @timofeev1995 in #312

New Contributors

• @nbroad1881 made their first contribution in #309
• @timofeev1995 made their first contribution in #312

Full Changelog: v1.5.7...1.5.8

v1.5.7

What's Changed

• Add a logic to support max_batch_size_dpo. by @VProv in #305

Full Changelog: v1.5.6...v1.5.7

v1.5.6

What's Changed

• Update README to show to use logprobs parameter by @nihit in #293
• fix(ft): download checkpoint query param by @alexnm in #294
• Tci file support by @zainhas in #301
• Update dependencies by @dependabot in #299
• Bump the dependencies group across 1 directory with 3 updates by @dependabot in #304

New Contributors

• @nihit made their first contribution in #293
• @alexnm made their first contribution in #294
• @zainhas made their first contribution in #301

Full Changelog: v1.5.5...v1.5.6

v1.5.5

What's Changed

• Add more information from response object to APIError during file upload by @avnermay in #289
• Fix empty LR scheduler for old jobs, clean up argument validation by @mryab in #288
• Make fine-tuning job validation messages more informative by @connermanuel in #292

New Contributors

• @avnermay made their first contribution in #289

Full Changelog: v1.5.4...v1.5.5

v1.5.4

What's Changed

• feat: code interpreter by @CompuIves in #263
• bump version to 1.5.4 by @orangetin in #285

Full Changelog: v1.5.3...v1.5.4