NanoCode012 842fa039dd feat: add sonicmoe fused lora support (#3519) ...

* feat: add sonicmoe fused lora support

* fix: forgot to add file

* feat: add test

* feat: add lora support for other routes

* fix: add int8 lora support

* fix: add qwen35_moe interleave support

* fix: qwen3_5_moe loss

* chore: lint

* address some pr comments

* fix test imports

* add support matrix for moe kernels [skip ci]

---------

Co-authored-by: Wing Lian <wing@axolotl.ai>

2026-04-02 08:53:48 -04:00

..

integrations

feat: add sonicmoe fused lora support (#3519)

2026-04-02 08:53:48 -04:00

kernels

post merge lora fixes for CI (#3536) [skip ci]

2026-03-23 02:26:10 -04:00

multigpu

upgrade transformers to 5.4.0 (#3562)

2026-03-31 19:15:59 -04:00

patched

feat: LoRA kernel support for bias, dropout, dora, embeddings (#3528) [skip ci]

2026-03-22 13:53:19 -04:00

solo

support flattening/packing for GRPO (#3552)

2026-03-28 13:15:54 -04:00

__init__.py

missing dunder-init

2023-11-06 18:33:01 -05:00

.gitignore

Support Sample packing for phi arch (#586)

2023-09-15 15:46:54 -04:00

test_activation_offloading.py

transformers v5 upgrade (#3272)

2026-01-27 17:08:24 -05:00

test_deepseekv3.py

fix token state json and mistral tokenizer issue (#3522) [skip ci]

2026-03-21 22:46:10 -04:00

test_diffusion.py

roundup_power2_divisions not needed with newer pytorch versions (#3540)

2026-03-24 15:40:05 -04:00

test_dpo.py

DPO transformers v0.29 fixes (#3560) [skip ci]

2026-03-31 19:04:53 -04:00

test_embeddings_lr.py

transformers v5 upgrade (#3272)

2026-01-27 17:08:24 -05:00

test_evaluate.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_falcon.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_gemma2.py

transformers v5 upgrade (#3272)

2026-01-27 17:08:24 -05:00

test_gemma3_text.py

transformers v5 upgrade (#3272)

2026-01-27 17:08:24 -05:00

test_imports.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_llama_pretrain.py

transformers v5 upgrade (#3272)

2026-01-27 17:08:24 -05:00

test_llama_vision.py

transformers v5 upgrade (#3272)

2026-01-27 17:08:24 -05:00

test_llama.py

automatically enable tf32 if supported (#3473) [skip ci]

2026-03-16 23:47:00 -04:00

test_load_model.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_lora_llama.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_mamba.py

transformers v5 upgrade (#3272)

2026-01-27 17:08:24 -05:00

test_mistral.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_mixtral.py

fix token state json and mistral tokenizer issue (#3522) [skip ci]

2026-03-21 22:46:10 -04:00

test_optimizers.py

fix token state json and mistral tokenizer issue (#3522) [skip ci]

2026-03-21 22:46:10 -04:00

test_packing_loss.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_phi.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_preprocess.py

Fix typos in some files (#3250) [skip ci]

2025-11-07 08:21:20 -05:00

test_process_reward_model_smollm2.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_profiler.py

allow profiling in mid-training rather from the start (#2899) [skip ci]

2025-07-14 20:11:11 -04:00

test_qat.py

add: support mxfp4 axo (#3375)

2026-03-05 13:40:45 -05:00

test_quantization.py

MX QAT patch (#3553)

2026-04-01 18:21:02 -04:00

test_qwen.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_reward_model_smollm2.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_save_first_step.py

transformers v5 upgrade (#3272)

2026-01-27 17:08:24 -05:00

test_schedulers.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_streaming.py

Changes from dataset_processes to dataset_num_proc (#3352) [skip ci]

2026-02-10 17:44:17 +07:00

test_tokenizer.py

add tokenizer_save_jinja_files to keep legacy behavior of including chat template in tokenizer_config.json (#3093)

2025-08-26 09:30:04 -04:00

utils.py

increase rtol, codecov informational only, don't silently fail errors w curl (#3534) [skip ci]

2026-03-22 13:54:03 -04:00