Wannabtl/modeltest

ChatTS-14B-timesense

This model is a fine-tuned version of /xll/models/ChatTS-14B on the generated_timeseries_bench_qa_new2, the generated_timeseries_bench_plus2, the tulu_ift, the model_detect_0823, the model_qa, the new_chatts_sft and the new_chatts_ift datasets.

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 1
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 32
• total_train_batch_size: 256
• total_eval_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.02
• training_steps: 900

Training results

Framework versions

• Transformers 4.52.4
• Pytorch 2.6.0+cu124
• Datasets 3.4.1
• Tokenizers 0.21.4