July-2025

## More VRAM reduction, faster & bug fixes

> Please update Unsloth! `pip install --upgrade --force-reinstall --no-deps --no-cache-dir unsloth unsloth_zoo`

1. **Gemma 3N Vision now works** and is fixed! Please re-download all model checkpoints (Unsloth will auto do it) Try [Kaggle Notebook](https://www.kaggle.com/code/danielhanchen/gemma-3n-4b-multimodal-finetuning-inference)! There is also a challenge with a **prize pool of $100,000!**
2. **Gemma 3** text and vision are all fixed for T4, and is much faster. Losses of 6 to 7 are now fixed - it should be 1 to 2.
3. **10 to 25% less VRAM consumption** for all models. Also faster compiling and less errors. Unsloth is now more stable!
4. **Downloads stuck at 90% to 95%** fixed!
5. **Qwen 2.5, Qwen 2, GLM** all fixed as well.
6. **GRPO** now works with latest main TRL
7. **Main TRL, PEFT, Transformers all work**
8. **Forced upgrading transformers** is now fixed.
9. **Falcon H1 finetuning** should work great! Notebooks incoming
10. **Devstral 1.1 and MedGemma 27B, 4B support with vision**
11. Many many many more bug fixes - this release of Unsloth should be much more stable and error tolerant!

> Please update Unsloth! `pip install --upgrade --force-reinstall --no-deps --no-cache-dir unsloth unsloth_zoo`

## What's Changed
* Gemma 3N by @danielhanchen in https://github.com/unslothai/unsloth/pull/2809
* Add instructions for installing unsloth on RTX 5090 by @jeromeku in https://github.com/unslothai/unsloth/pull/2812
* Add falcon h1 by @dhiaEddineRhaiem in https://github.com/unslothai/unsloth/pull/2650
* Granite4 support by @mmathew23 in https://github.com/unslothai/unsloth/pull/2799
* import undefined transformers_version for falcon model by @mmathew23 in https://github.com/unslothai/unsloth/pull/2822
* Fix LoftQ with FastBaseModel by @mehmetoguzderin in https://github.com/unslothai/unsloth/pull/2826
* Create stale.yml by @danielhanchen in https://github.com/unslothai/unsloth/pull/2832
* Create stale.yml by @danielhanchen in https://github.com/unslothai/unsloth/pull/2836
* Added conda/mamba section to blackwell installation readme by @rolandtannous in https://github.com/unslothai/unsloth/pull/2817
* Gemma 3N bug fixes by @danielhanchen in https://github.com/unslothai/unsloth/pull/2842
* Fix loftq None config for FastBaseModel by @mmathew23 in https://github.com/unslothai/unsloth/pull/2848
* Convert torch.bfloat16, torch.float16, etc. to vLLM valid dtypes by @rishabh135 in https://github.com/unslothai/unsloth/pull/2811
* [Feature] enable unsloth on amd gpu by @billishyahao in https://github.com/unslothai/unsloth/pull/2520
* Fix Gemma 3N by @danielhanchen in https://github.com/unslothai/unsloth/pull/2854
* fix quantized model parameter count method by @rolandtannous in https://github.com/unslothai/unsloth/pull/2855
* Update CSM for faster inference (no compile) by @mmathew23 in https://github.com/unslothai/unsloth/pull/2865
* Fix `UnslothTrainingArguments` not patching `trl.Config` properly by @Erland366 in https://github.com/unslothai/unsloth/pull/2873
* Fix unnecessary warning for transformers >= 4.53.0 by @mmathew23 in https://github.com/unslothai/unsloth/pull/2867
* Update README.md by @danielhanchen in https://github.com/unslothai/unsloth/pull/2885
* Many bug fixes by @danielhanchen in https://github.com/unslothai/unsloth/pull/2908
* silenty skip falcon h1 import if transformers_version < 4.53.0 by @mmathew23 in https://github.com/unslothai/unsloth/pull/2912
* Dynamically adjust get_per_token_logps [trl main upgrade] by @Datta0 in https://github.com/unslothai/unsloth/pull/2911
* [Intel] add intel gpu with vllm support by @leizhenyuan in https://github.com/unslothai/unsloth/pull/2903
* [bugs] fix for casual mask by @leizhenyuan in https://github.com/unslothai/unsloth/pull/2868
* Explicitly check if xformers exists for attention by @Datta0 in https://github.com/unslothai/unsloth/pull/2889
* Falcon H1: if mlp doesn't exist in layer module check for feed_forward by @mmathew23 in https://github.com/unslothai/unsloth/pull/2913
* Move inputs to right devices. by @Datta0 in https://github.com/unslothai/unsloth/pull/2919
* Many bug fixes by @danielhanchen in https://github.com/unslothai/unsloth/pull/2927

## New Contributors
* @dhiaEddineRhaiem made their first contribution in https://github.com/unslothai/unsloth/pull/2650
* @mehmetoguzderin made their first contribution in https://github.com/unslothai/unsloth/pull/2826
* @rishabh135 made their first contribution in https://github.com/unslothai/unsloth/pull/2811
* @billishyahao made their first contribution in https://github.com/unslothai/unsloth/pull/2520

**Full Changelog**: https://github.com/unslothai/unsloth/compare/June-2025...July-2025