Tag
Benchmarks
3essays ·All tags
Both Sides of Context Compression
Codemode-style tool gateways solved half of context compression. The other half — what happens to a tool's response after it returns — is still an open problem for the whole category. Here's a measured look at why it matters more than people think.
When Rubrics Become Rewards
A fair agent benchmark is not only a scoreboard. Two-arm runs with expert criteria produce the traces that become SFT, preference, and GRPO training signal — if you design the evaluation so the verdict is trustworthy.
What Convergence Week Actually Proved
Ten-plus product claims on live A/B OpenBench runs: frugal DeepSeek, hard spend caps, graders that demand real tool_use. On scores 1.0, off scores 0.0 — and every passing pack wraps an RTP reasoning trace the training flywheel can consume.