Issues

• ReferenceJudge: words in the judge's explanation override its leading yes/no verdict

  PrimeIntellect-ai/verifiers#2737 · Filter by author kv-248 opened on Oct 1, 2026
  1

• best-of-n: finalize() ranks a crashed (never-scored) sibling as reward 0, so it can be marked best over scored attempts and decide pass_at_n

  PrimeIntellect-ai/verifiers#2682 · Filter by author shaurya416 opened on Sep 25, 2026

• Chat streams: an upstream error event followed by [DONE] is recorded as a successful assistant turn

  PrimeIntellect-ai/verifiers#2659 · Filter by author Chaoqi31 opened on Sep 23, 2026

• Env-server pool never notices a dead worker; a worker that fails at startup hangs the eval

  PrimeIntellect-ai/verifiers#2571 · Filter by author willccbb opened on Sep 9, 2026

• Served runs rebuild every task as the taskset's declared Task type, silently dropping per-case subclasses

  PrimeIntellect-ai/verifiers#2570 · Filter by author willccbb opened on Sep 9, 2026

• Support Harbor [environment.env] and [environment.healthcheck]

  PrimeIntellect-ai/verifiers#2354 · Filter by author parkerpettit opened on Aug 13, 2026

• v1: how should Harbor tasks that provision services via ENTRYPOINT bring up the agent-side environment?

  PrimeIntellect-ai/verifiers#2321 · Filter by author SergeyArc opened on Aug 11, 2026
  3

• v1: apply agentic-judge task weights when an isolated solver fails

  PrimeIntellect-ai/verifiers#2197 · Filter by author hallerite opened on Jul 31, 2026
  2

• v1: harden capture_patch handling of pre-existing untracked filenames

  PrimeIntellect-ai/verifiers#2196 · Filter by author hallerite opened on Jul 31, 2026
  2

• v1: make isolated agentic-judge collect generic task artifacts and honor judge image

  PrimeIntellect-ai/verifiers#2195 · Filter by author hallerite opened on Jul 31, 2026

• v1: enforce artifact bounds during transfer and validate restore archives

  PrimeIntellect-ai/verifiers#2194 · Filter by author hallerite opened on Jul 31, 2026

• v1: model Harbor artifacts as durable outputs, not only grading transport

  PrimeIntellect-ai/verifiers#2189 · Filter by author hallerite opened on Jul 30, 2026

• Lean taskset: injected axiom bypasses the reward (#1271 #print axioms follow-up)

  PrimeIntellect-ai/verifiers#2177 · Filter by author dumko2001 opened on Jul 30, 2026
  1

• v1: adapt the meaning of concurrency for multi-agent envs

  PrimeIntellect-ai/verifiers#2052 · Filter by author hallerite opened on Jul 16, 2026

• Env server fails to start on Windows: ZMQ ipc:// transport is unsupported

  PrimeIntellect-ai/verifiers#2030 · Filter by author martian56 opened on Jul 15, 2026
  2

• Prometheus metrics logging from env server/worker

  PrimeIntellect-ai/verifiers#1188 · Filter by author mikasenghaas opened on Apr 19, 2026
  2

• Support getting each turn in a multi-turn task as single training sample

  PrimeIntellect-ai/verifiers#170 · Filter by author LuanBrt opened on Jul 28, 2025
  2