tongshuangwu
commited on
Commit
•
d1d2d2e
1
Parent(s):
80dcf88
add model
Browse files- README.md +59 -0
- config.json +33 -0
- merges.txt +0 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +1 -0
- tokenizer_config.json +1 -0
- training_args.bin +3 -0
- vocab.json +0 -0
README.md
ADDED
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language: "en"
|
3 |
+
tags:
|
4 |
+
- counterfactual generation
|
5 |
+
widget:
|
6 |
+
- text: "It is great for kids. <perturb> [negation] It [BLANK] great for kids. [SEP]"
|
7 |
+
---
|
8 |
+
|
9 |
+
# Erised
|
10 |
+
|
11 |
+
## Model description
|
12 |
+
|
13 |
+
This is a ported version of [Erised](https://homes.cs.washington.edu/~wtshuang/static/papers/2021-arxiv-polyjuice.pdf), the general-purpose counterfactual generator.
|
14 |
+
|
15 |
+
#### How to use
|
16 |
+
|
17 |
+
```python
|
18 |
+
from transformers import AutoTokenizer, AutoModelWithLMHead
|
19 |
+
|
20 |
+
tokenizer = AutoTokenizer.from_pretrained("uw-hai/jolyjuice")
|
21 |
+
model = AutoModelWithLMHead.from_pretrained("uw-hai/jolyjuice")
|
22 |
+
|
23 |
+
|
24 |
+
prompt_text = "It is great for kids. <perturb> [negation] It [BLANK] great for kids. [SEP]"
|
25 |
+
perturb_tok, end_tok = "<|perturb|>", "<|endoftext|>"
|
26 |
+
encoded_prompt = tokenizer.encode(prompt_text, add_special_tokens=False, return_tensors="pt")
|
27 |
+
input_ids = encoded_prompt
|
28 |
+
stop_token= '\n'
|
29 |
+
repetition_penalty=1
|
30 |
+
output_sequences = model.generate(
|
31 |
+
input_ids=input_ids,
|
32 |
+
max_length=100 + len(encoded_prompt[0]),
|
33 |
+
temperature=0.1,
|
34 |
+
num_beams=10,
|
35 |
+
num_return_sequences=3)
|
36 |
+
|
37 |
+
if len(output_sequences.shape) > 2:
|
38 |
+
output_sequences.squeeze_()
|
39 |
+
|
40 |
+
for generated_sequence_idx, generated_sequence in enumerate(output_sequences):
|
41 |
+
generated_sequence = generated_sequence.tolist()
|
42 |
+
# Decode text
|
43 |
+
text = tokenizer.decode(generated_sequence, clean_up_tokenization_spaces=True)
|
44 |
+
# Remove all text after the stop token
|
45 |
+
text = text[: text.find(stop_token) if stop_token and text.find(stop_token)>-1 else None]
|
46 |
+
text = text[: text.find(end_tok) if end_tok and text.find(end_tok)>-1 else None]
|
47 |
+
print(text)
|
48 |
+
```
|
49 |
+
|
50 |
+
### BibTeX entry and citation info
|
51 |
+
|
52 |
+
```bibtex
|
53 |
+
@article{wu2021polyjuice,
|
54 |
+
title={Polyjuice: Automated, General-purpose Counterfactual Generation},
|
55 |
+
author = {Wu, Tongshuang and Ribeiro, Marco Tulio and Heer, Jeffrey and Weld Daniel S.},
|
56 |
+
journal={arXiv preprint},
|
57 |
+
year={2021}
|
58 |
+
}
|
59 |
+
```
|
config.json
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"activation_function": "gelu_new",
|
3 |
+
"architectures": [
|
4 |
+
"GPT2LMHeadModel"
|
5 |
+
],
|
6 |
+
"attn_pdrop": 0.1,
|
7 |
+
"bos_token_id": 50256,
|
8 |
+
"embd_pdrop": 0.1,
|
9 |
+
"eos_token_id": 50256,
|
10 |
+
"gradient_checkpointing": false,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"layer_norm_epsilon": 1e-05,
|
13 |
+
"model_type": "gpt2",
|
14 |
+
"n_ctx": 1024,
|
15 |
+
"n_embd": 768,
|
16 |
+
"n_head": 12,
|
17 |
+
"n_inner": null,
|
18 |
+
"n_layer": 12,
|
19 |
+
"n_positions": 1024,
|
20 |
+
"resid_pdrop": 0.1,
|
21 |
+
"summary_activation": null,
|
22 |
+
"summary_first_dropout": 0.1,
|
23 |
+
"summary_proj_to_labels": true,
|
24 |
+
"summary_type": "cls_index",
|
25 |
+
"summary_use_proj": true,
|
26 |
+
"task_specific_params": {
|
27 |
+
"text-generation": {
|
28 |
+
"do_sample": true,
|
29 |
+
"max_length": 50
|
30 |
+
}
|
31 |
+
},
|
32 |
+
"vocab_size": 50257
|
33 |
+
}
|
merges.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bf8ed7c2e3b26b8a4d53e59f814b9fa03ae06d552ef08fc5f3f3bea4667e0406
|
3 |
+
size 510408315
|
special_tokens_map.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"bos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "eos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "unk_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}}
|
tokenizer_config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"bos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "eos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "unk_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "model_max_length": 1024}
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:043dbd5dd0744b2a7a925444abcf14b69fc74480e21f2eb163d4deb3381eba98
|
3 |
+
size 1839
|
vocab.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|