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
Downloads last month
116
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API: The model authors have turned it off explicitly.

Model tree for Walmart-the-bag/zephyr-quiklang-3b-4K

Finetuned
(1)
this model
Merges
1 model
Quantizations
2 models

Dataset used to train Walmart-the-bag/zephyr-quiklang-3b-4K

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