laion/openthoughts-4-code-qwen3-32b-annotated-32k_qwen3-1.7B_32k

Test Qwen3 1.7B post-trained on laion/openthoughts-4-code-qwen3-32b-32k-annotated (Qwen3 32B 32k output length annotated, using https://huggingface.co/datasets/mlfoundations-dev/herorun4_code as seed for generation)

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• —learning_rate: 0.00015
• —perdevicetrainbatchsize: 1
• —seed: 42
• —distributed_type: multi-GPU
• —num_devices: 256
• —totaltrainbatch_size: 256
• —optimizer: Use OptimizerNames.ADAMWTORCHFUSED with betas=(0.9,0.99) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• —lrschedulertype: cosine
• —lrschedulerwarmup_ratio: 0.1
• —num_epochs: 2.0

Training results

Framework versions

• —Transformers 4.55.0
• —Pytorch 2.7.1+cu128
• —Datasets 3.6.0
• —Tokenizers 0.21.1