File size: 1,402 Bytes
51581e2
3d684b2
 
51581e2
 
3d684b2
51581e2
3d684b2
 
51581e2
3d684b2
51581e2
 
 
 
3d684b2
51581e2
3d684b2
51581e2
3d684b2
 
 
 
 
51581e2
8f49427
51581e2
3d684b2
d1531a5
51581e2
 
 
3d684b2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
---
library_name: peft
base_model: google/flan-t5-large
---

# Model Card for honicky/t5-short-story-character-extractor

I trained this model as part of a learning project to build a children's story authoring tool for parents of young children.  See http://www.storytime.glass/
This model takes in a short story and outputs a comma separated list of characters in the story.

I'm not sure yet how useful this fine-tune is: rather it is for me to learn about the nuts and bolts of fine-tuning.


## Model Details

The model is a fine-tune of a sequence-to-sequence model [Flan T5 Large](https://huggingface.co./google/flan-t5-large), so a different architecture from decoder-only models like GPT*.  Maybe this allows it to perform this transformation task (transform a story into a list of characters) using a smaller model?  

 * Trained using `transformers.Seq2SeqTrainer` plus the corresponding collator, tokenizer etc	

- **Developed by:** RJ Honicky
- **Model type:** Encoder-Decoder Transformer
- **Language(s):** English (fine tune data set)
- **License:** MIT
- **Finetuned from model:** `google/flan-t5-large`

### Model Sources

- **Repository:** https://github.com/honicky/character-extraction
- **Weights and Biases**: https://wandb.ai/honicky/t5_target_finetune_for_character_extraction/runs/mx57gh45

## Uses

Primarily for use in https://github.com/honicky/story-time and for learning.