unsloth/unsloth
Daniel Han ca6c3dcc99 Train on responses only (#770)
* Update gemma2.py

* Update llama.py

* Update llama.py

* Update gemma2.py

* init

* Update gemma2.py

* Update gemma2.py

* Update _utils.py

* Update _utils.py

* Update _utils.py

* Update _utils.py

* Update _utils.py

* Update gemma2.py

* Update gemma2.py

* Update gemma2.py

* All RoPE Scaling support

* cleanup

* Update llama.py

* Update llama.py

* Update _utils.py

* Update _utils.py

* exec

* exec

* Attention_Module

* attention_module

* imports

* exec

* Update llama.py

* Update llama.py

* boolean mask

* revert masking

* Update llama.py

* Update save.py

* Update llama.py

* Update gemma2.py

* Update gemma2.py

* Update gemma2.py

* Update utils.py

* retry

* Update gemma2.py

* Update gemma2.py

* Update gemma2.py

* Update _utils.py

* Update _utils.py

* Update gemma2.py

* Update chat_templates.py

* Gemma 2 Ollama support

* Update llama.py

* Update llama.py

* error handling

* Update _utils.py

* Update _utils.py

* Stats for debugging

* Update _utils.py

* Update _utils.py

* Debugging

* Update tokenizer_utils.py

* Update _utils.py

* Update cross_entropy_loss.py

* Update cross_entropy_loss.py

* Update cross_entropy_loss.py

* Update rms_layernorm.py

* Update rms_layernorm.py

* Update rms_layernorm.py

* Update rms_layernorm.py

* Update rms_layernorm.py

* Update rms_layernorm.py

* Update rms_layernorm.py

* Check exec, eval

* Update _utils.py

* Update _utils.py

* Images

* Bug fixes

* Update pyproject.toml

* Bug fixes

* Update _utils.py

* Update _utils.py

* Deprecation fix

* Update chat_templates.py

* Now permitting use of pre-installed llama.cpp (#763)

* Now permitting use of pre-installed llama.cpp

* Update save.py

---------

Co-authored-by: Giuseppe Strafforello <giuseppe.strafforello@titantechnologies.com>
Co-authored-by: Daniel Han <danielhanchen@gmail.com>

* Update save.py

* Deprecation & compile

* typo

* Update chat_templates.py

* Update chat_templates.py

* train_on_responses_only

* Update llama.py

* Update llama.py

* Update save.py

* Update gemma2.py

---------

Co-authored-by: pepistrafforello <pepi.strafforello@gmail.com>
Co-authored-by: Giuseppe Strafforello <giuseppe.strafforello@titantechnologies.com>
2024-07-14 22:41:04 -07:00
..
kernels Train on responses only (#770) 2024-07-14 22:41:04 -07:00
models Train on responses only (#770) 2024-07-14 22:41:04 -07:00
__init__.py Debugging (#739) 2024-07-06 18:50:00 -07:00
chat_templates.py Train on responses only (#770) 2024-07-14 22:41:04 -07:00
save.py Train on responses only (#770) 2024-07-14 22:41:04 -07:00
tokenizer_utils.py Many bug fixes (#754) 2024-07-10 01:59:06 -07:00
trainer.py Fix is_bfloat16_supported missing (#510) 2024-05-22 20:40:57 +10:00