Nouamane/lighteval #356
Merged
Nouamane/lighteval #356
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…anotron into nouamane/lighteval
…into nouamane/lighteval
garongkim
pushed a commit
to garongkim/nanotron
that referenced
this pull request
Jun 2, 2025
* InitScalingMethod * InitScalingMethod * eval * try adding lightevalrunner to trainer * amend * amend * amend * amend * amend * amend * . * amend * amend * . * qos to low * add nanotron_path * some fix: logs, and config * cp instead of sync * eval_interval * serialize sanity checks * add output dir and s3_save path in the config * fix s3 only if define * fixes * add requeue * add wandb with lighteval and fix eval interval * fix this little space :( * folder_path should always have s3 when using s3 (fix consumed tokens issue) * config qwen * . --------- Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co>
NouamaneTazi
added a commit
that referenced
this pull request
Jun 23, 2025
* Nouamane/lighteval (#356) * InitScalingMethod * InitScalingMethod * eval * try adding lightevalrunner to trainer * amend * amend * amend * amend * amend * amend * . * amend * amend * . * qos to low * add nanotron_path * some fix: logs, and config * cp instead of sync * eval_interval * serialize sanity checks * add output dir and s3_save path in the config * fix s3 only if define * fixes * add requeue * add wandb with lighteval and fix eval interval * fix this little space :( * folder_path should always have s3 when using s3 (fix consumed tokens issue) * config qwen * . --------- Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co> * fix inference in case of varlen (input with paddings) * . * legacy * remove bos token * max-micro-batch * separate inference from training * use_decode_text * add no use cache case to decode_tokenized --------- Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co>
NouamaneTazi
added a commit
that referenced
this pull request
Jun 23, 2025
* can only merge to main from dev * Fix UnBoundLocalError in `clm_collator.py` (#339) * Update clm_collator.py * can only merge to main from dev (#348) --------- Co-authored-by: Nouamane Tazi <nouamane@huggingface.co> * fix init and init scaling factor and run evals in background (#349) * InitScalingMethod * InitScalingMethod * run evals in background (#352) * eval * try adding lightevalrunner to trainer * amend * amend * amend * amend * amend * amend * . * amend * amend * . * qos to low * add nanotron_path * some fix: logs, and config * cp instead of sync * eval_interval * serialize sanity checks * add output dir and s3_save path in the config * fix s3 only if define * fixes --------- Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co> --------- Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co> * [Feature] Implement CUDA event-based timing for improved GPU performa… (#346) * [Feature] Implement CUDA event-based timing for improved GPU performance measurement * can only merge to main from dev (#348) * Fix timer decorator logic: Support both CPU and CUDA timers and update docs * Fix timer decorator logic: support both CPU and CUDA; update docs --------- Co-authored-by: Kabir Grewal <kabirgrewal@Kabirs-MacBook-Pro-5.local> Co-authored-by: Nouamane Tazi <nouamane@huggingface.co> Co-authored-by: Kabir Grewal <kabirgrewal@kabirs-mbp-5.wireless-1x.unc.edu> * amend previous pr (#354) * MoE without token dropping (#355) * can only merge to main from dev (#348) * move moe from qwen modeling to src/nn * add groupedmlp * add token permute and unpermute * fix num_tokens_per_expert counting < num_experts * fix init and init scaling factor and run evals in background (#353) * can only merge to main from dev * Fix UnBoundLocalError in `clm_collator.py` (#339) * Update clm_collator.py * can only merge to main from dev (#348) --------- Co-authored-by: Nouamane Tazi <nouamane@huggingface.co> * fix init and init scaling factor and run evals in background (#349) * InitScalingMethod * InitScalingMethod * run evals in background (#352) * eval * try adding lightevalrunner to trainer * amend * amend * amend * amend * amend * amend * . * amend * amend * . * qos to low * add nanotron_path * some fix: logs, and config * cp instead of sync * eval_interval * serialize sanity checks * add output dir and s3_save path in the config * fix s3 only if define * fixes --------- Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co> --------- Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co> --------- Co-authored-by: Connector Switch <c8ef@outlook.com> Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co> * inference qwen moe seems to work inference seems good rn * update readme * fix router's weight initialization and wrong hidden size for non-moe mlp in qwen * add source for router weight and router logits in float32 * fixes * . * . * add parametrize grouped mlp in column and row linear * add logging per-param grad norm * fix conversation fail due to buffer on cpu * config_qwen * . * . * fix moe convert config --------- Co-authored-by: Nouamane Tazi <nouamane@huggingface.co> Co-authored-by: Connector Switch <c8ef@outlook.com> Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co> Co-authored-by: zzhhjjj <z785566960@gmail.com> Co-authored-by: nouamanetazi <nouamane98@gmail.com> * Nouamane/lighteval (#356) * InitScalingMethod * InitScalingMethod * eval * try adding lightevalrunner to trainer * amend * amend * amend * amend * amend * amend * . * amend * amend * . * qos to low * add nanotron_path * some fix: logs, and config * cp instead of sync * eval_interval * serialize sanity checks * add output dir and s3_save path in the config * fix s3 only if define * fixes * add requeue * add wandb with lighteval and fix eval interval * fix this little space :( * folder_path should always have s3 when using s3 (fix consumed tokens issue) * config qwen * . --------- Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co> * SmoLM3 training 🚀 (#375) * InitScalingMethod * InitScalingMethod * eval * try adding lightevalrunner to trainer * amend * amend * amend * amend * amend * amend * . * amend * amend * . * qos to low * add nanotron_path * some fix: logs, and config * cp instead of sync * eval_interval * serialize sanity checks * add output dir and s3_save path in the config * fix s3 only if define * fixes * add requeue * add wandb with lighteval and fix eval interval * fix this little space :( * folder_path should always have s3 when using s3 (fix consumed tokens issue) * fix resuming with new data mixture * offsets must be in samples not tokens * sanity check local files when dataset_read_path * better error for new stage * rmsnorm * sliding window * causal SWA * Revert "rmsnorm" This reverts commit 17dad0a. * rope_seq_len_interpolation_factor * logmixin for intermediate tensors + CP + consumed_token shenanigans when resuming training (#365) * logmixin * context parallelism (llama3 ring attn) + consumed_token shenanigans (#366) * training works * llama3 ring attn * llama3 ring attn * llama3 ring attn * fix position_ids (make them global) * rope_seq_len_interpolation_factor assert * . * . * fix rope and cp_pg * fixed consumed_tokens log --------- Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co> --------- Co-authored-by: Connector Switch <c8ef@outlook.com> Co-authored-by: elie <97572401+eliebak@users.noreply.github.com> Co-authored-by: “eliebak” <elie.bakouch@huggingface.co> Co-authored-by: grewalsk <136873529+grewalsk@users.noreply.github.com> Co-authored-by: Kabir Grewal <kabirgrewal@Kabirs-MacBook-Pro-5.local> Co-authored-by: Kabir Grewal <kabirgrewal@kabirs-mbp-5.wireless-1x.unc.edu> Co-authored-by: XλRI-U5 <b3f0cus@icloud.com> Co-authored-by: zzhhjjj <z785566960@gmail.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Fixes # (issue)
Before submitting
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.