ahans1 commited on
Commit
7ef90cd
1 Parent(s): 1d1a5d3

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +49 -182
README.md CHANGED
@@ -1,199 +1,66 @@
1
  ---
2
  library_name: transformers
3
- tags: []
 
 
 
 
 
 
 
 
 
4
  ---
5
 
6
- # Model Card for Model ID
7
 
8
- <!-- Provide a quick summary of what the model is/does. -->
 
9
 
 
10
 
 
 
 
11
 
12
- ## Model Details
 
 
13
 
14
- ### Model Description
15
 
16
- <!-- Provide a longer summary of what this model is. -->
17
 
18
- This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
 
 
 
 
 
 
 
 
19
 
20
- - **Developed by:** [More Information Needed]
21
- - **Funded by [optional]:** [More Information Needed]
22
- - **Shared by [optional]:** [More Information Needed]
23
- - **Model type:** [More Information Needed]
24
- - **Language(s) (NLP):** [More Information Needed]
25
- - **License:** [More Information Needed]
26
- - **Finetuned from model [optional]:** [More Information Needed]
27
 
28
- ### Model Sources [optional]
29
 
30
- <!-- Provide the basic links for the model. -->
 
31
 
32
- - **Repository:** [More Information Needed]
33
- - **Paper [optional]:** [More Information Needed]
34
- - **Demo [optional]:** [More Information Needed]
35
 
36
- ## Uses
37
 
38
- <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
39
-
40
- ### Direct Use
41
-
42
- <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
43
-
44
- [More Information Needed]
45
-
46
- ### Downstream Use [optional]
47
-
48
- <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
49
-
50
- [More Information Needed]
51
-
52
- ### Out-of-Scope Use
53
-
54
- <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
55
-
56
- [More Information Needed]
57
-
58
- ## Bias, Risks, and Limitations
59
-
60
- <!-- This section is meant to convey both technical and sociotechnical limitations. -->
61
-
62
- [More Information Needed]
63
-
64
- ### Recommendations
65
-
66
- <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
67
-
68
- Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
69
-
70
- ## How to Get Started with the Model
71
-
72
- Use the code below to get started with the model.
73
-
74
- [More Information Needed]
75
-
76
- ## Training Details
77
-
78
- ### Training Data
79
-
80
- <!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
81
-
82
- [More Information Needed]
83
-
84
- ### Training Procedure
85
-
86
- <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
87
-
88
- #### Preprocessing [optional]
89
-
90
- [More Information Needed]
91
-
92
-
93
- #### Training Hyperparameters
94
-
95
- - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
96
-
97
- #### Speeds, Sizes, Times [optional]
98
-
99
- <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
100
-
101
- [More Information Needed]
102
-
103
- ## Evaluation
104
-
105
- <!-- This section describes the evaluation protocols and provides the results. -->
106
-
107
- ### Testing Data, Factors & Metrics
108
-
109
- #### Testing Data
110
-
111
- <!-- This should link to a Dataset Card if possible. -->
112
-
113
- [More Information Needed]
114
-
115
- #### Factors
116
-
117
- <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
118
-
119
- [More Information Needed]
120
-
121
- #### Metrics
122
-
123
- <!-- These are the evaluation metrics being used, ideally with a description of why. -->
124
-
125
- [More Information Needed]
126
-
127
- ### Results
128
-
129
- [More Information Needed]
130
-
131
- #### Summary
132
-
133
-
134
-
135
- ## Model Examination [optional]
136
-
137
- <!-- Relevant interpretability work for the model goes here -->
138
-
139
- [More Information Needed]
140
-
141
- ## Environmental Impact
142
-
143
- <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
144
-
145
- Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
146
-
147
- - **Hardware Type:** [More Information Needed]
148
- - **Hours used:** [More Information Needed]
149
- - **Cloud Provider:** [More Information Needed]
150
- - **Compute Region:** [More Information Needed]
151
- - **Carbon Emitted:** [More Information Needed]
152
-
153
- ## Technical Specifications [optional]
154
-
155
- ### Model Architecture and Objective
156
-
157
- [More Information Needed]
158
-
159
- ### Compute Infrastructure
160
-
161
- [More Information Needed]
162
-
163
- #### Hardware
164
-
165
- [More Information Needed]
166
-
167
- #### Software
168
-
169
- [More Information Needed]
170
-
171
- ## Citation [optional]
172
-
173
- <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
174
-
175
- **BibTeX:**
176
-
177
- [More Information Needed]
178
-
179
- **APA:**
180
-
181
- [More Information Needed]
182
-
183
- ## Glossary [optional]
184
-
185
- <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
186
-
187
- [More Information Needed]
188
-
189
- ## More Information [optional]
190
-
191
- [More Information Needed]
192
-
193
- ## Model Card Authors [optional]
194
-
195
- [More Information Needed]
196
-
197
- ## Model Card Contact
198
-
199
- [More Information Needed]
 
1
  ---
2
  library_name: transformers
3
+ datasets:
4
+ - tomg-group-umd/wikipedia-en-2k-samples
5
+ tags:
6
+ - goldfish-loss
7
+ - memorization
8
+ - mitigation
9
+ license: apache-2.0
10
+ language:
11
+ - en
12
+ pipeline_tag: text2text-generation
13
  ---
14
 
15
+ # Quick Links
16
 
17
+ - **GitHub Repository**: https://github.com/ahans30/goldfish-loss
18
+ - **arXiv**: https://arxiv.org/abs/2406.10209
19
 
20
+ # Goldfish Loss
21
 
22
+ <div align="center">
23
+ <img src="https://raw.githubusercontent.com/ahans30/goldfish-loss/main/assets/goldfish-loss.jpg" width="300"/>
24
+ </div>
25
 
26
+ We introduce goldfish loss, a new language modeling loss function that mitigates memorization of training data.
27
+ Specifically, goldfish loss pseudorandomly drops $1/k$ of total tokens seen (in the forward pass) during loss computation (i.e., it doesn't compute loss for these tokens), with k being a hyperparameter.
28
+ We show that the model finds it increasingly difficult to verbatim regurgitate training data even after 100 epochs. Please read our paper linked below for more details.
29
 
30
+ # Overview
31
 
32
+ The following checkpoints are from our paper titled Goldfish Loss: Mitigating Memorization in Generative LLMs [[paper link](https://arxiv.org/abs/2406.10209)].
33
 
34
+ | Checkpoint Name | k-GL | Token Drop Strategy | Pretrain Tokens | Primary Dataset | Canaries Dataset for Memorization |
35
+ | ------------------------------------------------------------------------------------------------------------- | ---- | ------------------- | --------------- | --------------- | ----------------------------------------------------------------------------------- |
36
+ | [tomg-group-umd/3-goldfish-loss-llama-1B](https://huggingface.co/tomg-group-umd/3-goldfish-loss-llama-1B) | 3 | Hash (width = 13) | 20B | Redpajama | [Wikipedia](https://huggingface.co/datasets/tomg-group-umd/wikipedia-en-2k-samples) |
37
+ | [tomg-group-umd/4-goldfish-loss-llama-1B](https://huggingface.co/tomg-group-umd/4-goldfish-loss-llama-1B) | 4 | Hash (width = 13) | 20B | Redpajama | [Wikipedia](https://huggingface.co/datasets/tomg-group-umd/wikipedia-en-2k-samples) |
38
+ | [tomg-group-umd/8-goldfish-loss-llama-1B](https://huggingface.co/tomg-group-umd/8-goldfish-loss-llama-1B) | 8 | Hash (width = 13) | 20B | Redpajama | [Wikipedia](https://huggingface.co/datasets/tomg-group-umd/wikipedia-en-2k-samples) |
39
+ | [tomg-group-umd/32-goldfish-loss-llama-1B](https://huggingface.co/tomg-group-umd/32-goldfish-loss-llama-1B) | 32 | Hash (width = 13) | 20B | Redpajama | [Wikipedia](https://huggingface.co/datasets/tomg-group-umd/wikipedia-en-2k-samples) |
40
+ | [tomg-group-umd/128-goldfish-loss-llama-1B](https://huggingface.co/tomg-group-umd/128-goldfish-loss-llama-1B) | 128 | Hash (width = 13) | 20B | Redpajama | [Wikipedia](https://huggingface.co/datasets/tomg-group-umd/wikipedia-en-2k-samples) |
41
+ | [tomg-group-umd/control-llama-1B](https://huggingface.co/tomg-group-umd/control-llama-1B) | \- | No Tokens Dropped | 20B | Redpajama | None |
42
+ | [tomg-group-umd/standard-loss-llama-1B](https://huggingface.co/tomg-group-umd/standard-loss-llama-1B) | \- | No Tokens Dropped | 20B | Redpajama | [Wikipedia](https://huggingface.co/datasets/tomg-group-umd/wikipedia-en-2k-samples) |
43
 
44
+ ### Description
45
+ - `standard-loss-llama-1B` and `control-llama-1B` are trained with the standard causal language modeling loss, which has the same exact specifications as the goldfish models.
46
+ - The control model differs only in the fact that it did not utilize the canaries dataset for memorization and was simply pre-trained on 20B Redpajama tokens.
47
+ - The Canaries dataset, which contains 2000 Wikidocs, is repeated 50 times throughout the pre-training. Thus, it contains around ~204M tokens in total (including padding).
 
 
 
48
 
49
+ # Technical Specification
50
 
51
+ Each checkpoint mentioned above used randomly initialized [TinyLLaMA-1.1B](https://huggingface.co/TinyLlama/TinyLlama_v1.1) architecture.
52
+ For pretraining details, please find check our [GitHub](https://github.com/ahans30/goldfish-loss) repository.
53
 
54
+ # Cite our work
 
 
55
 
56
+ If you find our model, codebase or dataset beneficial, please consider citing our work:
57
 
58
+ ```bibtex
59
+ @misc{hans2024like,
60
+ title={Be like a Goldfish, Don't Memorize! Mitigating Memorization in Generative LLMs},
61
+ author={Abhimanyu Hans and Yuxin Wen and Neel Jain and John Kirchenbauer and Hamid Kazemi and Prajwal Singhania and Siddharth Singh and Gowthami Somepalli and Jonas Geiping and Abhinav Bhatele and Tom Goldstein},
62
+ year={2024},
63
+ eprint={2406.10209},
64
+ archivePrefix={arXiv},
65
+ }
66
+ ```