taiyi-lab

toolathlon-qwen36-27b-reasoning-sft-20260621-epoch1

Deploy Dedicated

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Get help setting up a custom Dedicated Endpoints.

Talk with our engineer to get a quote for reserved GPU instances with discounts.

README

License: other

Toolathlon Qwen3.6 27B Reasoning SFT - Epoch 1

This is epoch 1/3 from the local Toolathlon reasoning SFT full-finetune run.

Base model: /data/models/qwen3.6-27B
Local MCore model: /data/models/qwen3.6-27B-mcore
Local checkpoint at upload time: /data/train/toolathlon_qwen36_27b/output/gym5k_reasoning_20260621_full/reasoning_full_gym5k_3epoch/v0-20260621-091540/checkpoint-1058
Checkpoint step: 1058
Uploaded at: 2026-06-22 02:46:42 UTC
Visibility: public
Training framework: MS-Swift / Megatron, full fine-tune, bf16
Parallelism: tensor parallel size 8, pipeline parallel size 1
Training max length: 196,608
Dataset: taiyi-lab/toolathlon-gym-generated-param-5k-sft-20260618
Converted local training file: /data/train/toolathlon_qwen36_27b/data/gym5k_reasoning_train_192k.jsonl
Checkpoint size: 50.97 GiB
Safetensor shards: 12

The checkpoint is intended for Toolathlon self-hosted eval and research comparison against the base Qwen3.6 27B model and the other epochs from the same run.

Model provider

taiyi-lab

Model tree

Base

this model

Modalities

Input

Video, Text, Image

Output

Text

Pricing

Dedicated Endpoints