Rhythm Garg
@rhythmrg
Harvey's product harness and domain expertise is real alpha. It gives them something especially valuable for model training: high-quality, product-specific evals grounded in legal expertise.
For example, 20% of Harvey's inference spend was on closed-source models to power Review
For example, 20% of Harvey's inference spend was on closed-source models to power Review
Gabe Pereyra@gabepereyra · Aug 20Update on @harvey’s model training effort.
We post-trained a model we are calling Tenet:
- Achieves SOTA on LAB
- Generalizes 3rd party legal benchmarks
- Uses sub-agents for domain specific capabilities
Tenet uses Kimi K3 as base and was post-trained in collaboration
Open quoted post →We post-trained a model we are calling Tenet:
- Achieves SOTA on LAB
- Generalizes 3rd party legal benchmarks
- Uses sub-agents for domain specific capabilities
Tenet uses Kimi K3 as base and was post-trained in collaboration
0 35