2023

In-context Vectors: Making In Context Learning More Effective and Controllable Through Latent Space Steering

Liu, Sheng, Ye, Haotian, Xing, Lei et al.

Understand

Large language models (LLMs) demonstrate emergent in-context learning capabilities, where they adapt to new tasks based on example demonstrations.

However, in-context learning has seen limited effectiveness in many settings, is difficult to quantitatively control and takes up context window space.
To overcome these limitations, we propose an alternative approach that recasts in-context learning as in-context vectors (ICV).
Using ICV has two steps.

Built on

C.-Y. Lin, “Rouge: A package for automatic evaluation of summaries,” in
2004
Earlier work this paper cites.
W. Xu, A. Ritter, B. Dolan, R. Grishman, and C. Cherry, “Paraphrasing for style,” in
2012
Earlier work this paper cites.
X. Zhang, J. Zhao, and Y. LeCun, “Character-level convolutional networks for text classification,”
2015
Earlier work this paper cites.
T. Bolukbasi, K.-W. Chang, J. Y. Zou, V. Saligrama, and A. T. Kalai, “Man is to computer programmer as woman is to homemaker? debiasing word embeddings,”
2016
Earlier work this paper cites.
A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,”
2017
Earlier work this paper cites.
E. Wulczyn, N. Thain, and L. Dixon, “Ex machina: Personal attacks seen at scale,” in
2017
Earlier work this paper cites.
T. Shen, T. Lei, R. Barzilay, and T. Jaakkola, “Style transfer from non-parallel text by cross-alignment,”
2017
Earlier work this paper cites.
S. Rao and J. Tetreault, “Dear sir or madam, may i introduce the gyafc dataset: Corpus, benchmarks and metrics for formality style transfer,” in
2018
Earlier work this paper cites.
T. Zhang, V. Kishore, F. Wu, K. Q. Weinberger, and Y. Artzi, “Bertscore: Evaluating text generation with bert,” in
2019
Earlier work this paper cites.
T. Brown, B. Mann, N. Ryder, M. Subbiah, J. D. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell,
2020
Earlier work this paper cites.
D. Votipka, K. R. Fulton, J. Parker, M. Hou, M. L. Mazurek, and M. Hicks, “Understanding security mistakes developers make: Qualitative analysis from build it, break it, fix it,” in
2020
Earlier work this paper cites.
S. Gehman, S. Gururangan, M. Sap, Y. Choi, and N. A. Smith, “Realtoxicityprompts: Evaluating neural toxic degeneration in language models,” in
2020
Earlier work this paper cites.
S. Gururangan, A. Marasović, S. Swayamdipta, K. Lo, I. Beltagy, D. Downey, and N. A. Smith, “Don’t stop pretraining: Adapt language models to domains and tasks,” in
2020
Earlier work this paper cites.
A. Conneau, K. Khandelwal, N. Goyal, V. Chaudhary, G. Wenzek, F. Guzmán, É. Grave, M. Ott, L. Zettlemoyer, and V. Stoyanov, “Unsupervised cross-lingual representation learning at scale,” in
2020
Earlier work this paper cites.
M. Skjuve, A. Følstad, K. I. Fostervold, and P. B. Brandtzaeg, “My chatbot companion-a study of human-chatbot relationships,”
2021
Earlier work this paper cites.
A. Holtzman, P. West, V. Shwartz, Y. Choi, and L. Zettlemoyer, “Surface form competition: Why the highest probability answer isn’t always right,” in
2021
Earlier work this paper cites.
E. J. Hu, Y. Shen, P. Wallis, Z. Allen-Zhu, Y. Li, S. Wang, L. Wang, and W. Chen, “Lora: Low-rank adaptation of large language models,”
Original
2021
Earlier work this paper cites.
J. Xu, D. Ju, M. Li, Y.-L. Boureau, J. Weston, and E. Dinan, “Bot-adversarial dialogue for safe conversational agents,” in
2021
Earlier work this paper cites.
C. Guo, A. Sablayrolles, H. Jégou, and D. Kiela, “Gradient-based adversarial attacks against text transformers,” in
2021
Earlier work this paper cites.
S. M. Xie, A. Raghunathan, P. Liang, and T. Ma, “An explanation of in-context learning as implicit bayesian inference,” in
2021
Earlier work this paper cites.
T. Schick, S. Udupa, and H. Schütze, “Self-diagnosis and self-debiasing: A proposal for reducing corpus-based bias in nlp,”
2021
Earlier work this paper cites.
P. Riley, N. Constant, M. Guo, G. Kumar, D. C. Uthus, and Z. Parekh, “Textsettr: Few-shot text style extraction and tunable targeted restyling,” in
2021
Earlier work this paper cites.
S. Black, S. Biderman, E. Hallahan, Q. Anthony, L. Gao, L. Golding, H. He, C. Leahy, K. McDonell, J. Phang,
2022
Earlier work this paper cites.

Similar

Q. Dong, L. Li, D. Dai, C. Zheng, Z. Wu, B. Chang, X. Sun, J. Xu, and Z. Sui, “A survey for in-context learning,”
Original
2022
Cited alongside, same era.
S. Min, X. Lyu, A. Holtzman, M. Artetxe, M. Lewis, H. Hajishirzi, and L. Zettlemoyer, “Rethinking the role of demonstrations: What makes in-context learning work?,” in
2022
Cited alongside, same era.
J. Liu, D. Shen, Y. Zhang, W. B. Dolan, L. Carin, and W. Chen, “What makes good in-context examples for gpt-3?,” in
2022
Cited alongside, same era.
V. Logacheva, D. Dementieva, S. Ustyantsev, D. Moskovskiy, D. Dale, I. Krotova, N. Semenov, and A. Panchenko, “Paradetox: Detoxification with parallel data,” in
2022
Cited alongside, same era.
O. Rubin, J. Herzig, and J. Berant, “Learning to retrieve prompts for in-context learning,” in
2022
Cited alongside, same era.
S. Ye, D. Kim, J. Jang, J. Shin, and M. Seo, “Guess the instruction! flipped learning makes language models stronger zero-shot learners,”
Original
2022
Cited alongside, same era.
S. Min, M. Lewis, H. Hajishirzi, and L. Zettlemoyer, “Noisy channel language model prompting for few-shot text classification,” in
2022
Cited alongside, same era.
B. Xu, Q. Wang, Z. Mao, Y. Lyu, Q. She, and Y. Zhang, “k nn prompting: Beyond-context learning with calibration-free nearest neighbor inference,” in
2022
Cited alongside, same era.
C. Burns, H. Ye, D. Klein, and J. Steinhardt, “Discovering latent knowledge in language models without supervision,”
Original
2022
Cited alongside, same era.
K. Li, A. K. Hopkins, D. Bau, F. Viégas, H. Pfister, and M. Wattenberg, “Emergent world representations: Exploring a sequence model trained on a synthetic task,”
Original
2022
Cited alongside, same era.
Y. Lu, M. Bartolo, A. Moore, S. Riedel, and P. Stenetorp, “Fantastically ordered prompts and where to find them: Overcoming few-shot prompt order sensitivity,” in
2022
Cited alongside, same era.
S. Shin, S. W. Lee, H. Ahn, S. Kim, H. S. Kim, B. Kim, K. Cho, G. Lee, W. Park, J. W. Ha,
2022
Cited alongside, same era.

Then

A. Zou, Z. Wang, J. Z. Kolter, and M. Fredrikson, “Universal and transferable adversarial attacks on aligned language models, 2023,”
2023
Closest in time.
L. Zheng, W.-L. Chiang, Y. Sheng, S. Zhuang, Z. Wu, Y. Zhuang, Z. Lin, Z. Li, D. Li, E. Xing,
Original
2023
Closest in time.
H. Touvron, T. Lavril, G. Izacard, X. Martinet, M.-A. Lachaux, T. Lacroix, B. Rozière, N. Goyal, E. Hambro, F. Azhar,
Original
2023
Closest in time.
G. Penedo, Q. Malartic, D. Hesslow, R. Cojocaru, A. Cappelli, H. Alobeidli, B. Pannier, E. Almazrouei, and J. Launay, “The refinedweb dataset for falcon llm: outperforming curated corpora with web data, and web data only,”
Original
2023
Closest in time.
W.-L. Chiang, Z. Li, Z. Lin, Y. Sheng, Z. Wu, H. Zhang, L. Zheng, S. Zhuang, Y. Zhuang, J. E. Gonzalez, I. Stoica, and E. P. Xing, “Vicuna: An open-source chatbot impressing gpt-4 with 90%* chatgpt quality,” March 2023
2023
Closest in time.
Y. Wen, N. Jain, J. Kirchenbauer, M. Goldblum, J. Geiping, and T. Goldstein, “Hard prompts made easy: Gradient-based discrete optimization for prompt tuning and discovery,” in
2023
Closest in time.
G. Ilharco, M. T. Ribeiro, M. Wortsman, L. Schmidt, H. Hajishirzi, and A. Farhadi, “Editing models with task arithmetic,” in
2023
Closest in time.
F. Yin, J. Vig, P. Laban, S. Joty, C. Xiong, and C.-S. J. Wu, “Did you read the instructions? rethinking the effectiveness of task definitions in instruction learning,”
Original
2023
Closest in time.
X. Wan, R. Sun, H. Nakhost, H. Dai, J. M. Eisenschlos, S. O. Arik, and T. Pfister, “Universal self-adaptive prompting,”
Original
2023
Closest in time.
X. Wan, R. Sun, H. Dai, S. O. Arik, and T. Pfister, “Better zero-shot reasoning with self-adaptive prompting,”
Original
2023
Closest in time.
J. Yang, B. Hui, M. Yang, B. Li, F. Huang, and Y. Li, “Iterative forward tuning boosts in-context learning in language models,”
Original
2023
Closest in time.
G. Cui, W. Li, N. Ding, L. Huang, Z. Liu, and M. Sun, “Decoder tuning: Efficient language understanding as decoding,” in
2023
Closest in time.
R. Hendel, M. Geva, and A. Globerson, “In-context learning creates task vectors,”
Original
2023
Closest in time.
A. Turner, L. Thiergart, D. Udell, G. Leech, U. Mini, and M. MacDiarmid, “Activation addition: Steering language models without optimization,”
Original
2023
Closest in time.
A. Zou, L. Phan, S. Chen, J. Campbell, P. Guo, R. Ren, A. Pan, X. Yin, M. Mazeika, A.-K. Dombrowski,
Original
2023
Closest in time.
U. Mini, P. Grietzer, M. Sharma, A. Meek, M. MacDiarmid, and A. M. Turner, “Understanding and controlling a maze-solving policy network,”
Original
2023
Closest in time.
K. Li, O. Patel, F. Viégas, H. Pfister, and M. Wattenberg, “Inference-time intervention: Eliciting truthful answers from a language model,”
Original
2023
Closest in time.
J. Wei, J. Wei, Y. Tay, D. Tran, A. Webson, Y. Lu, X. Chen, H. Liu, D. Huang, D. Zhou,
Original
2023
Closest in time.
J. Von Oswald, E. Niklasson, E. Randazzo, J. Sacramento, A. Mordvintsev, A. Zhmoginov, and M. Vladymyrov, “Transformers learn in-context by gradient descent,” in
2023
Closest in time.
F. Bianchi, M. Suzgun, G. Attanasio, P. Röttger, D. Jurafsky, T. Hashimoto, and J. Zou, “Safety-tuned llamas: Lessons from improving the safety of large language models that follow instructions,”
Original
2023
Closest in time.
C. T. Leong, Y. Cheng, J. Wang, J. Wang, and W. Li, “Self-detoxifying language models via toxification reversal,”
Original
2023
Closest in time.
N. Meade, S. Gella, D. Hazarika, P. Gupta, D. Jin, S. Reddy, Y. Liu, and D. Hakkani-Tür, “Using in-context learning to improve dialogue safety,”
Original
2023
Closest in time.
A. Som, K. Sikka, H. Gent, A. Divakaran, A. Kathol, and D. Vergyri, “Demonstrations are all you need: Advancing offensive content paraphrasing using in-context learning,”
Original
2023
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…

Reading the bibliography…