Commit Graph

  • b76d2d1130 Update: review comments! Rahul Tuli 2025-03-26 21:57:00 +00:00
  • 7946f89df4 Add: SFTPlugin with llmcompressor Rahul Tuli 2025-03-12 07:09:06 +00:00
  • 5dbfa3ef1f Built site for gh-pages Quarto GHA Workflow Runner 2025-04-28 16:21:02 +00:00
  • 1178a15ede Feat: Add qwen3 and CCE for qwen family (#2518) NanoCode012 2025-04-28 23:18:46 +07:00
  • 51c5ba0276 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-28 16:14:25 +00:00
  • c513487d1a support val_set_size for splitting test split from train with DPO (#2572) Wing Lian 2025-04-28 12:12:15 -04:00
  • 3f2fbb75b1 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-28 15:22:54 +00:00
  • dda95e6c40 add preview-docs workflow (#2432) Dan Saunders 2025-04-28 11:20:46 -04:00
  • 14b3af3330 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-28 14:13:30 +00:00
  • 7099343c56 feat: add eos_tokens and train_on_eot for chat_template EOT parsing (#2364) NanoCode012 2025-04-28 21:11:20 +07:00
  • 5000cb3fe7 grab sys prompt too from dataset (#2397) [skip ci] Wing Lian 2025-04-28 10:11:06 -04:00
  • 170cdb5be9 Add Post_model_load, post_lora_load, post_train, post_train_unload function calls (#2539) divyanshuaggarwal 2025-04-28 19:40:28 +05:30
  • 5d182a1056 Add runpod sls handler (#2530) [skip ci] Ezekiel Wotring 2025-04-28 06:08:32 -08:00
  • 40f4ea23ab replace references to random 68m model w 135m smollm2 (#2570) [skip ci] Wing Lian 2025-04-28 10:08:07 -04:00
  • f1df73a798 fix(doc): clarify vllm usage with grpo (#2573) [skip ci] NanoCode012 2025-04-28 21:07:45 +07:00
  • d48bc7afb6 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-28 04:34:18 +00:00
  • 8b33ae1c4f Fix bug in grpo reward module import (#2571) Dhruv Mullick 2025-04-27 22:31:56 -06:00
  • 19280435f5 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-27 23:22:12 +00:00
  • dc4da4a7e2 update trl to 0.17.0 (#2560) Wing Lian 2025-04-27 19:19:53 -04:00
  • 2b9a2dde4b chore: update title runpod-sls NanoCode012 2025-04-25 15:24:33 +07:00
  • 388e950016 restore dockerfile Wing Lian 2025-04-24 13:11:00 -04:00
  • fb4adbb311 fix: trim allowed cuda versions NanoCode012 2025-04-24 11:22:27 +07:00
  • 5e8abca54f use axolotl cloud image as base and various fixes Wing Lian 2025-04-23 13:22:17 -04:00
  • 168ec339e5 chore: lint Wing Lian 2025-04-22 14:13:48 -04:00
  • cb7185998b remove LICENSE and fix README zeke 2025-04-14 18:33:27 -08:00
  • c2fc35f520 Add runpod sls handler zeke 2025-04-14 18:30:14 -08:00
  • 5251facdd0 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-26 18:17:05 +00:00
  • f9c7c3bb72 don't use is_main_process during config validation (#2569) Wing Lian 2025-04-26 14:14:52 -04:00
  • b708a1cc45 validate config to set defaults llmcompressor-sft Wing Lian 2025-04-26 13:11:25 -04:00
  • 35a2679e12 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-26 01:13:30 +00:00
  • caf5cb63ea add e2e smoke test for using activation/gradient checkpointing with offload (#2565) Wing Lian 2025-04-25 21:11:17 -04:00
  • 5dba5c82a8 fix support for wandb run_name for rl trainers (#2566) [skip ci] Wing Lian 2025-04-25 21:10:54 -04:00
  • 1a5e1f477d Built site for gh-pages Quarto GHA Workflow Runner 2025-04-25 21:17:16 +00:00
  • e3c9d541a7 fix: crash when pretraining_dataset with dispatch_batches is false (#2558) Chiwan Park 2025-04-26 06:15:03 +09:00
  • 9eba0ad118 chore(doc): update docker tags on doc (#2559) [skip ci] NanoCode012 2025-04-26 04:14:48 +07:00
  • 53dbf97d85 make cce default to true when using the plugin (#2562) [skip ci] Wing Lian 2025-04-25 17:14:26 -04:00
  • 2c2563bc34 fix: gradient checkpointing functools.partial object has no attribute __self__ (#2563) [skip ci] Eko Julianto Salim 2025-04-26 04:02:37 +07:00
  • 5cb3398460 don't fail on codecov upload for external contributor PRs (#2564) [skip ci] Wing Lian 2025-04-25 15:10:55 -04:00
  • 9f52387e0d Built site for gh-pages Quarto GHA Workflow Runner 2025-04-25 14:36:13 +00:00
  • ae1c7ace63 Sequence parallel training context manager (#2553) Dan Saunders 2025-04-25 10:33:54 -04:00
  • 926dc4af90 updates sp-rl-v3 Dan Saunders 2025-04-25 02:28:38 +00:00
  • 6810f0ee19 minimize diffs to GRPO trainer Dan Saunders 2025-04-23 19:04:26 +00:00
  • 6c65eeaaf7 finalizing SP + GRPO trainer Dan Saunders 2025-04-22 03:38:46 +00:00
  • 7f4e4076e1 progress Dan Saunders 2025-04-18 21:35:33 +00:00
  • 4f2d092216 subclassing constructor Dan Saunders 2025-04-17 18:37:22 +00:00
  • b13b6e185f stronger subclassing of TRL GRPO trainer; custom distributed sampler Dan Saunders 2025-04-17 04:06:18 +00:00
  • 76e2d2e60b progress Dan Saunders 2025-04-14 21:02:30 +00:00
  • 11b6803ff4 grpo sp support Dan Saunders 2025-04-09 00:46:05 +00:00
  • e55dce9995 fix Dan Saunders 2025-04-16 14:07:09 +00:00
  • 9640aacfc9 fixes for batch API funcs, simplify Dan Saunders 2025-04-16 03:47:51 +00:00
  • 5306c6acbb fix Dan Saunders 2025-04-14 14:41:52 +00:00
  • 4ae8df16a9 adding all batch ring-flash-attn methods via single adapter Dan Saunders 2025-04-11 05:08:08 +00:00
  • 74e7cfd28f update Dan Saunders 2025-04-11 04:06:37 +00:00
  • 2bb5c1fe7e batch api HF adapter for ring-flash-attn; cleanup and improvements Dan Saunders 2025-04-11 03:45:34 +00:00
  • 3f1873cc62 pytest Dan Saunders 2025-04-24 16:19:47 +00:00
  • 072df89e0e add gather post hook, simplify, fixes Dan Saunders 2025-04-24 14:10:03 +00:00
  • cb7c3ee847 tweak codecov yaml Dan Saunders 2025-04-24 00:20:05 +00:00
  • d92ac7a41d reorg Dan Saunders 2025-04-24 00:11:37 +00:00
  • 5816433121 nit Dan Saunders 2025-04-24 00:02:40 +00:00
  • e5a4e21497 simplifying Dan Saunders 2025-04-23 23:56:31 +00:00
  • 65ae78009c simplifying Dan Saunders 2025-04-23 23:49:11 +00:00
  • 7e5168ad74 accommodate both training context managers Dan Saunders 2025-04-23 23:40:45 +00:00
  • cd393fecc3 further simplifying Dan Saunders 2025-04-23 23:37:41 +00:00
  • bac5568bda update Dan Saunders 2025-04-23 23:30:47 +00:00
  • 69aeae80ed updates Dan Saunders 2025-04-23 23:19:52 +00:00
  • cafda804ec ctx manager for SP Dan Saunders 2025-04-23 19:49:37 +00:00
  • daa9a58f83 Add: line about further optimizations using llmcompressor Rahul Tuli 2025-04-24 14:06:25 -04:00
  • ae7069e15b Merge branch 'main' into llmcompressor-sft Rahul Tuli 2025-04-24 12:37:14 -05:00
  • 20d48cd617 Address Review Comments: * deleted redundant docs/llm_compressor.qmd * incorporated feedback in integration README.md * added llmcompressor integration to docs/custom_integrations.qmd Rahul Tuli 2025-04-23 18:00:00 -04:00
  • 49fac6d310 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-24 17:04:03 +00:00
  • 1447beb132 make sure to validate the config before normalizing so defaults get set (#2554) merged-2554 Wing Lian 2025-04-24 13:01:43 -04:00
  • e766a730ba Add: .qmd file Rahul Tuli 2025-04-21 20:40:49 -04:00
  • 7dc797860e Tests, Style, Updates Rahul Tuli 2025-04-21 20:33:59 -04:00
  • ff4904c8c4 Rebase and updates! Rahul Tuli 2025-04-17 17:19:59 -04:00
  • 45b7293793 Add: llm_compressor integration documentation Rahul Tuli 2025-04-09 01:03:45 +00:00
  • 279c7178bc Move: LLMCompressorPlugin into it's own submodule Rahul Tuli 2025-04-09 00:27:48 +00:00
  • e73c3709f9 Update model config Rahul Tuli 2025-04-08 23:53:29 +00:00
  • 33562189f8 Use: absolute import Rahul Tuli 2025-04-08 23:51:49 +00:00
  • c057a2268f Rename: sft.yaml to sparse-finetuning.yaml Rahul Tuli 2025-04-08 23:46:32 +00:00
  • 9d7a3809b5 Add: llcompressor installable Rahul Tuli 2025-04-08 23:35:12 +00:00
  • b7b24d6a64 Address review comments from @markurtz Rahul Tuli 2025-04-04 17:59:41 +00:00
  • 8b82b8f7a1 Apply suggestions from @markurtz Rahul Tuli 2025-04-04 10:36:35 -04:00
  • 81da58c0a1 Update llmcompressor version to latest Rahul Tuli 2025-04-03 09:37:43 -04:00
  • 2cd5a234a7 Revert: TODO's Rahul Tuli 2025-04-02 22:54:22 +00:00
  • 8c1af0747d Use: warning over warn Rahul Tuli 2025-04-02 22:38:31 +00:00
  • a06b360d99 pre commit hooks Rahul Tuli 2025-04-02 22:35:36 +00:00
  • 0f6456a14f Add:llmcompressor instalable Rahul Tuli 2025-04-02 22:22:04 +00:00
  • 47a333ce49 Update: review comments! Rahul Tuli 2025-03-26 21:57:00 +00:00
  • f9d6776c28 Add: SFTPlugin with llmcompressor Rahul Tuli 2025-03-12 07:09:06 +00:00
  • 3179a36e87 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-24 12:54:15 +00:00
  • 66f41ec6f1 disable codecov pr annotations (#2556) Dan Saunders 2025-04-24 08:51:51 -04:00
  • 8a645a9541 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-24 05:05:55 +00:00
  • 85053f4bd4 Fix(doc): add delinearize instruction (#2545) NanoCode012 2025-04-24 12:03:43 +07:00
  • 0812992467 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-24 04:42:03 +00:00
  • a4d5112ae1 builds for torch 2.7.0 (#2552) Wing Lian 2025-04-24 00:39:31 -04:00
  • 4f64594182 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-23 19:01:19 +00:00
  • 0d691cc2a7 add base docker image with pytorch 2.7.0 and variant for cuda 12.8 (#2551) Wing Lian 2025-04-23 14:59:03 -04:00
  • 872acc75b3 Built site for gh-pages Quarto GHA Workflow Runner 2025-04-23 14:36:00 +00:00
  • c4053481ff Codecov fixes / improvements (#2549) Dan Saunders 2025-04-23 10:33:30 -04:00
  • caa234bfdf Built site for gh-pages Quarto GHA Workflow Runner 2025-04-23 14:30:22 +00:00