conda CI

Workflow runs for conda CI

• Support fully async coding-agent RL with 128k Qwen contexts

  skipped:
  conda CI#4432:Pull request#2474synchronizebyzhuzilin
  zilin/coding-agent

  Oct 15, 20263s
  More actions for workflow run 4432

• Support fully async coding-agent RL with 128k Qwen contexts

  skipped:
  conda CI#4431:Pull request#2474synchronizebyzhuzilin
  zilin/coding-agent

  Oct 15, 20262s
  More actions for workflow run 4431

• Support fully async coding-agent RL with 128k Qwen contexts

  skipped:
  conda CI#4430:Pull request#2474openedbyzhuzilin
  zilin/coding-agent

  Oct 15, 20263s
  More actions for workflow run 4430

• fix(coding_agent_rl): snapshot fails when repo .gitignore ignores .harness

  skipped:
  conda CI#4429:Pull request#2473openedbydextermayhewjd
  dextermayhewjd:fix/coding-agent-snapshot-ignored-harness

  Oct 15, 20261s
  More actions for workflow run 4429

• Add Qwen3-30B FP8-experts deterministic stack (DeepGEMM, residual checkpoint, sampling seeds)

  skipped:
  conda CI#4428:Pull request#2472openedbyfy1214
  fy1214:pr/deterministic-qwen3-core

  Oct 15, 20261s
  More actions for workflow run 4428

• fix(agent): restore a re-rendered assistant echo before rendering the prompt

  skipped:
  conda CI#4427:Pull request#2282synchronizebypollyloop
  pollyloop:fix/agent-tool-call-echo-drops-turn

  Oct 13, 20267s
  More actions for workflow run 4427

• test: disable git auto-maintenance in swe workspace diff test

  skipped:
  conda CI#4426:Pull request#2469openedbygavinqii
  gavinqii:fix/swe-workspace-diff-maintenance-race

  Oct 13, 20269s
  More actions for workflow run 4426

• Feat/prism omni colocate rl

  skipped:
  conda CI#4425:Pull request#2468synchronizebykaysonyu
  kaysonyu:feat/prism-omni-colocate-rl

  Oct 12, 20261s
  More actions for workflow run 4425

• fix(qwen3.5): average TP-replicated linear-attention gradients across TP

  skipped:
  conda CI#4424:Pull request#2451synchronizebygavinqii
  gavinqii:fix/qwen3.5-tp-replicated-grad-sync

  Oct 12, 20261s
  More actions for workflow run 4424

• Feat/prism omni colocate rl

  skipped:
  conda CI#4423:Pull request#2468openedbykaysonyu
  kaysonyu:feat/prism-omni-colocate-rl

  Oct 12, 20261s
  More actions for workflow run 4423

• fix(tau-bench): train on sampled token ids instead of retokenized chat text

  skipped:
  conda CI#4422:Pull request#2467openedbyPlanckChang
  PlanckChang:fix/tau-bench-token-exact

  Oct 12, 20262s
  More actions for workflow run 4422

• fix(tau-bench): train on sampled token ids instead of retokenized chat text

  skipped:
  conda CI#4421:Pull request#2466openedbyPlanckChang
  PlanckChang:fix/tau-bench-sampled-tokens

  Oct 12, 20261s
  More actions for workflow run 4421

• [docs] Refine index page

  skipped:
  conda CI#4420:Pull request#2465openedbyzhuzilin
  docs

  Oct 12, 20267s
  More actions for workflow run 4420

• [docs] update index page

  skipped:
  conda CI#4419:Pull request#2464openedbyzhuzilin
  docs

  Oct 12, 20262s
  More actions for workflow run 4419

• Add opt-in R3 route-mismatch metric and targeted R3 CI checks

  skipped:
  conda CI#4418:Pull request#2463openedbykar137
  kar137:r3-route-mismatch

  Oct 11, 20262s
  More actions for workflow run 4418

• fix(data): preserve empty datasets during prompt length filtering

  skipped:
  conda CI#4417:Pull request#2462openedbyGoldArowana
  GoldArowana:fix/empty-dataset-prompt-length-20261011

  Oct 11, 20261s
  More actions for workflow run 4417

• perf: fuse FP32 CUDA updates in StatelessAdam

  skipped:
  conda CI#4416:Pull request#2461openedbyCharlesXu-HQ
  CharlesXu-HQ:perf/fused-stateless-adam

  Oct 11, 20261s
  More actions for workflow run 4416

• fix(ray): bind trainer CPU affinity to the selected GPU

  skipped:
  conda CI#4415:Pull request#2460openedbyDongDongJu
  DongDongJu:fix/trainer-cpu-affinity

  Oct 11, 20261s
  More actions for workflow run 4415

• test: prevent Git maintenance race in SWE workspace fixture

  skipped:
  conda CI#4414:Pull request#2459openedbyGoldArowana
  GoldArowana:fix/swe-fixture-disable-auto-maintenance

  Oct 10, 20261s
  More actions for workflow run 4414

• fix(qwen3.5): average TP-replicated linear-attention gradients across TP

  skipped:
  conda CI#4413:Pull request#2451synchronizebygavinqii
  gavinqii:fix/qwen3.5-tp-replicated-grad-sync

  Oct 10, 20267s
  More actions for workflow run 4413

• perf(agent): isolate prompt rendering from event loop

  skipped:
  conda CI#4412:Pull request#2328synchronizebyliuhao-labs
  liuhao-labs:codex/2285-tokenizer-worker

  Oct 10, 20261s
  More actions for workflow run 4412

• fix(megatron): compact actor logits before projection

  skipped:
  conda CI#4411:Pull request#2310synchronizebyliuhao-labs
  liuhao-labs:fix/2253-compact-actor-logits

  Oct 10, 20261s
  More actions for workflow run 4411

• Add sunabako agent E2E and align top-p score centering

  skipped:
  conda CI#4410:Pull request#2457synchronizebyzhuzilin
  zilin/agent-e2e-text

  Oct 10, 20266s
  More actions for workflow run 4410

• Add sunabako agent E2E and align top-p score centering

  skipped:
  conda CI#4409:Pull request#2457synchronizebyzhuzilin
  zilin/agent-e2e-text

  Oct 10, 20269s
  More actions for workflow run 4409

• Add sunabako agent E2E and align top-p score centering

  skipped:
  conda CI#4408:Pull request#2457synchronizebyzhuzilin
  zilin/agent-e2e-text

  Oct 10, 20261s
  More actions for workflow run 4408