Downloads · 30 days
5
3% of all-time downloads
sonthenguyen/zephyr-sft-bnb-4bit-DPO-dismissive_mtbc-252steps
zephyr-sft-bnb-4bit-DPO-dismissive_mtbc-252steps is a text generation model from sonthenguyen. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as apache-2.0.
TrainOutput(globalstep=252, trainingloss=0.0682461416144461, metrics={'trainruntime': 2060.2317, 'trainsamplespersecond': 1.953, 'trainstepspersecond': 0.122, 'totalflos': 0.0, 'trainloss': 0.0682461416144461, 'epoch'…
Downloads · 30 days
5
3% of all-time downloads
All-time downloads
168
Public
Parameters
7.2B
14.5 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors14.5 GB · 100%
From the Hugging Face model README
TrainOutput(global_step=252, training_loss=0.0682461416144461, metrics={'train_runtime': 2060.2317, 'train_samples_per_second': 1.953, 'train_steps_per_second': 0.122, 'total_flos': 0.0, 'train_loss': 0.0682461416144461, 'epoch': 0.5009940357852882}) ==((====))== Unsloth - 2x faster free finetuning | Num GPUs = 1 \ /| Num examples = 8,046 | Num Epochs = 1 O^O/ _/ \ Batch size per device = 4 | Gradient Accumulation steps = 4 \ / Total batch size = 16 | Total steps = 252 "-____-" Number of trainable parameters = 41,943,040
This mistral model was trained 2x faster with Unsloth and Huggingface's TRL library.