A 2026 summer research intern at Jane Street, Kavish, developed a diffusion model aimed at synthesizing order book events—such as order placements, cancellations and executions—rather than just predicting prices. The project explored whether market data, which mixes discrete actions with near-continuous and often spiky distributions (like order timing and price ticks), can be modeled using diffusion techniques borrowed from image, video, and audio generation.
blog.janestreet.com
· 2026-10-09
A developer published Jevgpt, a hobby project posted on Show HN, describing it as an autoregressive chatbot built using classifier-based small token vocabularies. The creator says they addressed limited token exploration by adding a small speculative decoding pass, using the same model as both drafter and verifier to expand vocabulary options. The code is available on GitHub under the user giga-james.
news.ycombinator.com
· 2026-09-23
AMD published results from experiments running speculative decoding in vLLM on its MI300X and MI355X Instinct GPUs using the ROCm software stack. The team compared five drafting approaches—native MTP, Gemma 4 MTP, EAGLE-3, DFlash, and DSpark—which differ in how draft tokens are generated and how they receive signals from the target model.
vllm.ai
· 2026-09-07
A technical essay argues that describing large language models as mere next-token predictors is outdated once reinforcement learning with verifiable rewards (RLVR) enters the picture. Unlike pre-training, which only reinforces sequences already present in training data, RLVR lets models generate novel token sequences and learn from evaluating their outcomes. The piece walks through pseudocode contrasting the two training loops to show how post-training changes what the model is actually optimizing for.
gmcgoldr.github.io
· 2026-09-04
A technical walkthrough details how diffusion language models work as an alternative to the dominant autoregressive approach used by most LLMs today. Rather than generating text token-by-token in sequence, diffusion models produce an entire sequence at once and refine it over multiple steps, drawing on techniques like masking, iterative refinement, and post-training used in recent open-source models. The material stems from workshop talks and lectures given at ICLR 2026 and MLSS 2026.
kuleshov-group.github.io
· 2026-08-30
A wave of new research papers is reviving interest in continuous diffusion techniques for language generation, an approach that had largely been overtaken by discrete diffusion and autoregressive methods in recent years. The author reviews this resurgence, tracing why continuous diffusion for text fell out of favor and why several research groups are now revisiting it as an alternative to token-by-token autoregressive generation.
sander.ai
· 2026-08-30
GoKawiil