Julius Cheng
@juliuscheng.bsky.social
Finishing up PhD in NLP at University of Cambridge. Deciding whether to put my weirdo ML thoughts on here or just be normal
They say it's something like 20-30% but 0% of my papers get accepted!! Something definitely wrong here
January 23, 2025 at 11:05 AM
They say it's something like 20-30% but 0% of my papers get accepted!! Something definitely wrong here
Our experiments are on machine translation, but this method works with any generator + reranker setup!
Eager to hear your thoughts, and happy reranking!
Eager to hear your thoughts, and happy reranking!
January 23, 2025 at 1:32 AM
Our experiments are on machine translation, but this method works with any generator + reranker setup!
Eager to hear your thoughts, and happy reranking!
Eager to hear your thoughts, and happy reranking!
Bonus: we show how to use multi-fidelity Bayesian optimization to use a smaller and faster proxy scoring model to search even more efficiently. We get the best performance by training a distilled model from our main CometKiwi model.
January 23, 2025 at 1:32 AM
Bonus: we show how to use multi-fidelity Bayesian optimization to use a smaller and faster proxy scoring model to search even more efficiently. We get the best performance by training a distilled model from our main CometKiwi model.
The candidate pool is actually a search space, and you can model your uncertainty about scores you haven't scored yet with GP regression. Use BayesOpt to search the pool for promising candidates.
This nearly gets the maximum achievable score with only 70/200 scoring calls!
This nearly gets the maximum achievable score with only 70/200 scoring calls!
January 23, 2025 at 1:32 AM
The candidate pool is actually a search space, and you can model your uncertainty about scores you haven't scored yet with GP regression. Use BayesOpt to search the pool for promising candidates.
This nearly gets the maximum achievable score with only 70/200 scoring calls!
This nearly gets the maximum achievable score with only 70/200 scoring calls!
Reranking is expensive and we show that you don't need to score every candidate in the candidate pool.
Use Bayesian optimization with GPs!
Use Bayesian optimization with GPs!
January 23, 2025 at 1:32 AM
Reranking is expensive and we show that you don't need to score every candidate in the candidate pool.
Use Bayesian optimization with GPs!
Use Bayesian optimization with GPs!
Language models for MT are good at generating large candidate pools that contain good translations; they're less good at assigning the highest score to the best translation.
This is where reranking comes in: rescoring with COMET, noisy channel decoding, minimum Bayes risk, etc.
This is where reranking comes in: rescoring with COMET, noisy channel decoding, minimum Bayes risk, etc.
January 23, 2025 at 1:32 AM
Language models for MT are good at generating large candidate pools that contain good translations; they're less good at assigning the highest score to the best translation.
This is where reranking comes in: rescoring with COMET, noisy channel decoding, minimum Bayes risk, etc.
This is where reranking comes in: rescoring with COMET, noisy channel decoding, minimum Bayes risk, etc.