AfterQuery Post-Trains Nemotron 3.5 Super VL: Finance Accuracy 47%→74% at 1/7 the Cost
ycombinator · x · 2026-10-16
Y Combinator retweeted AfterQuery's case study: the team post-trained NVIDIA's Nemotron 3.5 Super VL on expert credit tasks with Hebbia:
- Perfect-solved rate jumped from 47% to 74%, beating Sol 6
- Lands within 3 points of frontier model Astra 6 at roughly 1/7 the cost
- Takeaway: domain post-training can push open models past the cost-accuracy frontier on specialized tasks
Full case study linked in the original thread.
More from Research
- Anthropic's GRAFT routes flagged gradients into a deletable adapter, cutting reward hacking to near zero — rohanpaul_ai · 2026-10-16
- Someone found a hilariously specific 'betrayed by your own side' neuron in Qwen3-32B — NathanpmYoung · 2026-10-16
- "Decompiling Neural Networks into Code" Claim Draws Fire Over Its Theory — fkasummer · 2026-10-16
- Burkov's AI weekly #182: State of AI 2026, AI protein design, 260M-param diffusion beats 6x larger baselines — burkov · 2026-10-16
- Google demos magic state cultivation on Willow, cutting errors 40x with 99.99% fidelity — sundarpichai · 2026-10-16
- Interpretability paper incoming: in-depth look at testing explanations on modern LLMs — burny_tech · 2026-10-16