Walmart-the-bag/zephyr-quiklang-3b-4K

Description

This is the 4K version of https://huggingface.co./Walmart-the-bag/zephyr-quiklang-3b with 1000 more samples of openhermes.

Original Model Description

This is a finetune of StableLM-Zephyr-3B with 2 datasets, toxic-dpo and openhermes with 10000 samples.

Training Parameters

• 1xA6000-48GB
• batch_size: 6
• learning_rate: 5e-5

Datasets:

• unalignment/toxic-dpo-v0.1
• teknium/openhermes

Metrics/Basic Eval:

"predict_bleu-4": 31.594154999999997,
"predict_rouge-1": 44.092935,
"predict_rouge-2": 22.276081000000005,
"predict_rouge-l": 34.506909,
"predict_runtime": 121.7549,
"predict_samples_per_second": 0.821,
"predict_steps_per_second": 0.107