The State-Space/Mamba-370M is finetuned on ROC Stories dataset to be able to generate endings to short stories cohesively.

The Evaluation metrics on the ROC stories dataset for story ending generation are:

Bert (f1) : 0.878

Meteor: 0.1

bleu : 0.0125

Rouge1: 0.18

Perplexity : 207

To use the Model:

>>> from transformers import MambaForCausalLM, AutoTokenizer
>>> model_name = "DdIiVvYyAaMm/mamba-370m-story-generation"

>>> tokenizer = AutoTokenizer.from_pretrained(model_name)
>>> model = MambaForCausalLM.from_pretrained(model_name)

# And the rest of code standard as from transformers library.
Downloads last month
23
Safetensors
Model size
159M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.