Predacons

What's Changed
* Update requirements.txt and setup.py, add new functions by shouryashashank in https://github.com/Predacons/predacons/pull/20
* Update setup.py by shouryashashank in https://github.com/Predacons/predacons/pull/21

- moved to SFTTrainer
- added auto tokenizer
- added auto peft config using laura

- [ ] With this pr we can now train bigger model with very less memory
- [ ] tested training gemma with 4gb vram

**Full Changelog**: https://github.com/Predacons/predacons/compare/v0.0.116...v0.0.117

What's Changed
* optimized performance for generate operation by shouryashashank in https://github.com/Predacons/predacons/pull/14


What's Changed
* optimized performance for generate operation by shouryashashank in https://github.com/Predacons/predacons/pull/14


**Full Changelog**: https://github.com/Predacons/predacons/compare/v0.0.114...v0.0.115

What's Changed
* Feature/speculative fast generation by shouryashashank in https://github.com/Predacons/predacons/pull/13
* adds ability to generate upto 6 times faster response compared to native methods from hugging face and pytorch, https://arxiv.org/pdf/2211.17192.pdf
* gpt fast feature inspired and copied from https://github.com/armed-gpt/gpt-blazing , https://pytorch.org/blog/accelerating-generative-ai-2/, https://github.com/pytorch-labs/gpt-fast, https://github.com/MDK8888/GPTFast

**Full Changelog**: https://github.com/Predacons/predacons/compare/v0.0.113...v0.0.114

What's Changed
* added try catch for checkpoint by shouryashashank in https://github.com/Predacons/predacons/pull/12


**Full Changelog**: https://github.com/Predacons/predacons/compare/v0.0.112...v0.0.113

What's Changed
* added resume feature in train by shouryashashank in https://github.com/Predacons/predacons/pull/6
* bug fix in generate_text_data_source_llm by shouryashashank in https://github.com/Predacons/predacons/pull/7



**Full Changelog**: https://github.com/Predacons/predacons/compare/v0.0.109...v0.0.112