Show HN: Tier – Adaptive tool routing that makes small LLMs 10pt more accurate
1–6 of 6 posts
Re: Show HN: Tier – Adaptive tool routing that makes small LLMs 10pt more accurate
#2[dead]
Re: Show HN: Tier – Adaptive tool routing that makes small LLMs 10pt more accurate
#3[flagged]
Re: Show HN: Tier – Adaptive tool routing that makes small LLMs 10pt more accurate
#4[flagged]
Fixed eval — 80 tools, 200 queries, 4 model sizes. +10pp came from "all tools" vs "tiered" on 1.5B.
You're right about stability. Haven't run rotated/rephrased evals yet. The 89% baseline (when models knew where to look) suggests selection capability is fine, but I'd expect some regression with adversarial prompts.
Re: Show HN: Tier – Adaptive tool routing that makes small LLMs 10pt more accurate
#5This is cool. I am working on something similar for code schema reads by AI, which cost me a lot of tokens. I’ll share once battle tested. The idea of abstracting and then giving it a tree to follow is where I landed also.
Re: Show HN: Tier – Adaptive tool routing that makes small LLMs 10pt more accurate
#6This is cool. I am working on something similar for code schema reads by AI, which cost me a lot of tokens. I’ll share once battle tested. The idea of abstracting and then giving it a tree to follow is where I landed also.
Let me know your findings.