RichardErkhov/shi-zheng-qxhs_-_gpt2_oasst2_curated-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

gpt2oasst2curated - GGUF

• —Model creator: https://huggingface.co/shi-zheng-qxhs/
• —Original model: https://huggingface.co/shi-zheng-qxhs/gpt2oasst2curated/

Original model description: --- license: mit base_model: gpt2 tags:

• —generatedfromtrainer model-index:
• —name: gpt2oasst2curated results: [] datasets:
• —sablo/oasst2_curated language:
• —en pipeline_tag: text-generation widget:
• —text: Who is the president of the United States?
• —text: Hi, my name is Superman. How are you!
• —text: Do you know the history of Chelsea Football Club? inference: parameters: maxlength: 128 penaltyalpha: 0.6 topk: 6 padtokenid: 50256 eostokenid: 50256 libraryname: transformers ---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

gpt2oasst2curated

This model is a fine-tuned version of gpt2 on sablo/oasst2_curated dataset

Model description

For experimental purpose - chatbot

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: 0.0005
• —trainbatchsize: 32
• —evalbatchsize: 32
• —seed: 42
• —gradientaccumulationsteps: 8
• —totaltrainbatch_size: 256
• —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• —lrschedulertype: cosine
• —lrschedulerwarmup_steps: 1000
• —num_epochs: 20
• —mixedprecisiontraining: Native AMP

Training results

Framework versions

• —Transformers 4.36.2
• —Pytorch 2.0.0
• —Datasets 2.16.1
• —Tokenizers 0.15.0