Search results

Request to implement SCALE optimizer

Status: Open.

#2928 In pytorch/torchtune;

· jim-plus opened on Sep 17, 2025

Bug when I run from a checkpoint full parameter fine-tuning on single GPU

Status: Open.

#2927 In pytorch/torchtune;

· lilitmovsesian opened on Sep 16, 2025

support packing with flash_attn v2?

Status: Open.

#2925 In pytorch/torchtune;

· iqiancheng opened on Sep 9, 2025

Qwen 3 0.6B full_finetune_distributed breaks with context parallelism enabled

Status: Open.

#2924 In pytorch/torchtune;

· jerryyiransun opened on Sep 6, 2025

Can we distill from baseline teacher to baseline student without doing LoRA-tuning of student?

Status: Open.

#2922 In pytorch/torchtune;

· asmitamohanty opened on Aug 31, 2025

Lora Qwen-MoE adapter_model.pt merge to base model weight

Status: Open.

#2920 In pytorch/torchtune;

· iqiancheng opened on Aug 28, 2025

H2D takes large time overhead

Status: Open.

#2917 In pytorch/torchtune;

· iqiancheng opened on Aug 22, 2025

How to train Qwen3 with W_per_channel_Sym-A_per_token_Sym？

Status: Open.

#2916 In pytorch/torchtune;

· taishan1994 opened on Aug 19, 2025

DistributedCheckpointer argument conflicts

Status: Open.

#2907 In pytorch/torchtune;

· jdwillard19 opened on Aug 6, 2025

Support gpt-oss-120b and gpt-oss-20b

Status: Open.

#2903 In pytorch/torchtune;

· SLR722 opened on Aug 5, 2025

Support OLMo-2-1124-7B?

Status: Open.

#2902 In pytorch/torchtune;

· yangdongchao opened on Aug 5, 2025

Key error exception when finetuning Qwen models with LoRA

Status: Open.

#2895 In pytorch/torchtune;

· nouranali opened on Jul 30, 2025