chain-of-thought — Having a model write out intermediate reasoning steps before its answer, which markedly improves performance on hard problems.
test-time compute — Spending more computation when the model answers — thinking longer, trying multiple attempts — to buy accuracy without training a bigger model.
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
Why it matters
A carefully sourced deep dive into why test-time computeSpending more computation when the model answers — thinking longer, trying multiple attempts — to buy accuracy without training a bigger model.Full definition → and chain-of-thoughtHaving a model write out intermediate reasoning steps before its answer, which markedly improves performance on hard problems.Full definition → improve LLMA large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.Full definition → reasoning, giving engineers a mental model for when and how to spend inference-time compute effectively.
Key quotes
“In Transformer models, the amount of computation (flops) that the model does for each generated token is roughly 2 times the number of parameters. For sparse models like mixture of experts (MoE), only a fraction of the parameters are used in each forward pass, so computation = 2 * parameters / sparsity, where sparsity is the fraction of experts active.”
Lilian Weng
“Snell et al. (2024 ) showed that easier questions benefit from purely sequential test-time compute, whereas harder questions often perform best with an optimal ratio of sequential to parallel compute.”
Lilian Weng
“Interestingly the DeepSeek team showed that with pure RL, no SFT stage, it is still possible to learn advanced reasoning capabilities like reflection and backtracking (“Aha moment”). The model naturally learns to spend more thinking tokens during the RL training process to solve reasoning tasks.”
Lilian Weng
“They failed to use process reward model (PRM) as it is hard to define per-step rubrics or determine whether an intermediate step is correct, meanwhile making the training more vulnerable to reward hacking.”
Lilian Weng
“Chain of thought provides an especially convenient form of interpretability, as it makes the model’s internal process visible in natural language. This interpretability, however, rests on the assumption that the model truthfully describes its internal thought processes.”
articleWhen Chain-of-Thought Helps and When It Hurts: An Empirical Investigation of the Serial-Depth Bottleneck in LLM ReasoningTughanbulut Kurtulush
articleCategories of Inference-Time Scaling for Improved LLM ReasoningSebastian Raschka, PhD
articleScaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model ParametersCharlie Snell et al.