Harvey后训练Nemotron 3.5 Lightning，法律Agent基准得分8.3%
https://t.co/H0II4NxEnz
Harvey把Nemotron 3.5 Lightning后训练了一把，法律Agent评测从0冲到8.3%，比Opus 4.6还强，输出也从90k缩到37k，省Token。
Harvey联合Trajectory Labs对NVIDIA的Nemotron 3.5 Lightning进行后训练，在Legal Agent Bench上得分从0%升至8.
3%。该成绩超过了Opus 4.
6和更大的后训练版Nemotron 3 Ultra。后训练覆盖九个执业领域，未出现能力退化。
平均输出长度从90k tokens降至37k tokens，每token奖励提升2.4倍。
https://t.co/H0II4NxEnz
x.com/harvey/status/… Harvey @harvey We post-trained @NVIDIAAI Nemotron 3.5 Lightning on Legal Agent Bench with @trajectorylabs . Here's what we found: 1) Post-training improved agent performance from 0% to 8.3% on held-out LAB tasks, beating both Opus 4.6 and the much larger post-trained Nemotron 3 Ultra. 2) Performance improved across nine practice areas with no regressions. 3) Post-training reduced average model output from 90k to 37k tokens, increasing the model's reward-per-token by 2.4x. Through our collaboration with NVIDIA and Trajectory we’re committed to pushing the frontier of legal intelligence and cost efficiency with open weight models. Deep dive: 🔗 View Quoted Tweet
This report is categorized under AITOP's "Models" desk. Key entities and technical topics under continuous tracking include Nemotron 3.5 Lightning, NVIDIA, Harvey, Legal Agent Bench.
Source attribution: curated from NVIDIA AI (@NVIDIAAI). Cross-source corroboration: 11 independent sources have covered this development (including ollama, Aravind Srinivas, Fireworks AI), confirming industry-wide signal strength.
