Task gallery / Machine learning / Version 1
Task evidence / console
squeeze-qwen3.5-0.8b-300mb
uncalibratedChosen benchmark version- Reference solution (oracle)
- Unverified
- No-op resistance check
- Unverified
- QA verdict
- Not recorded
- Aggregate eligibility
- Eligible version
The task
Squeeze a 0.8B language model into 300 MB without wrecking it The situation We want to ship the text model inside Qwen3.5-0.8B-Base on devices that give us about 300 MB for the weights. In bf16 the model is 1.5 GB (752 million parameters), so we need roughly a 5x reduction, and the naive way of getting there is ugly: rounding every weight to 3 bits more than triples the model's perplexity. Your job is to find a much…
Excerpt only; not a complete executable task specification. Task ID task_16ce94f5b938473e · version 1. Selection status: not fully recorded.
Results by model and effort
This exact task version only. Every rate is passes / valid runs with a Wilson 95% interval; n≤2 remains counts only. Macro aggregation does not apply within a single task. Baselines are summarized in the quality gates, not ranked. A passed no-op resistance check means that doing nothing did not solve the task.
Current evidence
| Model / effort | Verifier result [95% Wilson] | Diagnostic labels / valid | HARNESS_ERROR · excluded | Outcome classes |
|---|---|---|---|---|
| GPT-5.6-sol · default (unverified) | 0/0 | 0 / 0 | 9 | GOOD_FAILURE: 0 BAD_FAILURE: 0 GOOD_SUCCESS: 0 BAD_SUCCESS: 0 |
| Claude Opus 5 · default (unverified) | 0/0 | 0 / 0 | 11 | GOOD_FAILURE: 0 BAD_FAILURE: 0 GOOD_SUCCESS: 0 BAD_SUCCESS: 0 |
Legacy evidence
| Model / effort | Verifier result [95% Wilson] | Diagnostic labels / valid | HARNESS_ERROR · excluded | Outcome classes |
|---|
Every recorded run
20 recorded trials, including controls. Diagnostic labels do not replace verifier reward. Labels and quality checks are source declarations, not independently attested here. Internal evidence remains private; a missing public URL is shown explicitly. Records with invalid or absent rewards are not scored.
| Run / model | Date / source | Reward / eligibility | Diagnostic class / source | Resources | Evidence |
|---|---|---|---|---|---|
trial_7261256a6ab329b4GPT-5.6-sol · default (unverified) | Sep 4, 2026 console | 0 · excludedharness_error | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_7981264c55ceeab0Claude Opus 5 · default (unverified) | Sep 4, 2026 console | 0 · excludedharness_error | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_ab0d1266c6a22ca6Claude Opus 5 · default (unverified) | Sep 4, 2026 console | 0 · excludedharness_error | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_74e2c92e59181f3fGPT-5.6-sol · default (unverified) | Sep 4, 2026 console | 0 · excludedharness_error | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_4dfa0da74b04f1a8GPT-5.6-sol · default (unverified) | Sep 5, 2026 console | 0 · excludedharness_error | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_6b67205a5a8d7129Claude Opus 5 · default (unverified) | Sep 5, 2026 console | 0 · excludedharness_error | HARNESS_ERROR derived | Tokens: 248,957 Turns: 300 Cost: $48.9953 | Not publicly available |
trial_bbd66ff69c3f9ed1Claude Opus 5 · default (unverified) | Sep 5, 2026 console | 0 · excludedharness_error | HARNESS_ERROR derived | Tokens: 131,314 Turns: 84 Cost: $10.7622 | Not publicly available |
trial_da5ba59619ff093dGPT-5.6-sol · default (unverified) | Sep 5, 2026 console | 0 · excludedharness_error | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_c7cb1cd2d61ae34cGPT-5.6-sol · default (unverified) | Sep 5, 2026 console | No reward · excludedno_reward | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_1c55e2b97ef47cf3GPT-5.6-sol · default (unverified) | Sep 5, 2026 console | No reward · excludedno_reward | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_7c1b104ce0996bc6GPT-5.6-sol · default (unverified) | Sep 5, 2026 console | No reward · excludedno_reward | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_9867491f63eb050dGPT-5.6-sol · default (unverified) | Sep 5, 2026 console | No reward · excludedno_reward | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_88f33db98651e2f1Claude Opus 5 · default (unverified) | Sep 5, 2026 console | 0 · excludedharness_error | HARNESS_ERROR derived | Tokens: 203,506 Turns: 315 Cost: $42.2144 | Not publicly available |
trial_725ceba6e91baa74GPT-5.6-sol · default (unverified) | Sep 5, 2026 console | No reward · excludedno_reward | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_aa9756849167b225Claude Opus 5 · default (unverified) | Sep 5, 2026 console | 0 · excludedharness_error | HARNESS_ERROR derived | Tokens: — Turns: — Cost: — | Not publicly available |
trial_c223419e43dd2de1Claude Opus 5 · default (unverified) | Sep 5, 2026 console | No reward · excludedno_reward | HARNESS_ERROR derived | Tokens: 0 Turns: 1 Cost: $0.0000 | Not publicly available |
trial_721b7100d3eaece9Claude Opus 5 · default (unverified) | Sep 5, 2026 console | No reward · excludedno_reward | HARNESS_ERROR derived | Tokens: 0 Turns: 1 Cost: $0.0000 | Not publicly available |
trial_fc3dc6f6d9990f08Claude Opus 5 · default (unverified) | Sep 5, 2026 console | No reward · excludedno_reward | HARNESS_ERROR derived | Tokens: 0 Turns: 1 Cost: $0.0000 | Not publicly available |
trial_bc8968da6f63b914Claude Opus 5 · default (unverified) | Sep 5, 2026 console | No reward · excludedno_reward | HARNESS_ERROR derived | Tokens: 0 Turns: 1 Cost: $0.0000 | Not publicly available |
trial_3aa67ee4c6d256eaClaude Opus 5 · default (unverified) | Sep 5, 2026 console | No reward · excludedno_reward | HARNESS_ERROR derived | Tokens: 0 Turns: 1 Cost: $0.0000 | Not publicly available |