Steve0927/qwen3-embedding-0.6b-lrat-full80k-hardneg-preserve-mr2-min6-from-broad
This is a Qwen3-Embedding-0.6B compatible dense retriever checkpoint prepared for XIR competition 1170.
Method
The method is Trajectory Utility Calibration (TUC).
We train only from official LRAT trajectory-derived retrieval rows. LRAT converts agent search/browse/reasoning trajectories into query-positive-negative training examples. TUC adds a conservative second-pass row-internal audit:
- identify weak positives;
- identify high-confidence false-negative negatives;
- apply bounded sample weights;
- preserve enough hard negatives for contrastive training.
Data Boundary
Training data source:
- data/processed/lrat_broad_g16_v2_llm_full80k_hardneg_preserve_mr2_min6_weights.jsonl
Base checkpoint:
- models/qwen3-embedding-0.6b-lrat-broad-g16-v2
Compliance boundary:
- No generated queries.
- No generated answers.
- No generated passages or external evidence.
- No A/B leaderboard qrels or hidden failures are used as training labels.
- Strong models are used only as auditors over official LRAT query-positive-negative rows.
Training Configuration
- Learning rate:
3e-7 - Epochs:
0.5 - Group size:
16 - Batch per device:
24 - Thresholds:
drop_low_pos_quality<=1.0; false_negative relevance>=3.5 confidence>=0.75; max_remove_per_row=2; min_neg=6; weight=[0.4,2.0]
Gate Evidence
Full80k gate assessment:
- Verdict:
pass - Report:
reports/full80k_gate_assessment.json
Dense gate:
- Report:
reports/dense_gate_full80k_hardneg_preserve_mr2_min6_vs_full80k_conservative.json - Summary:
pass vs broad_g16_v2 and pass vs full80k_conservative; recall@50 +0.001185 vs current best; hit@50 flat
Intended Use
This checkpoint is intended as a dense retriever in the XIR/BrowseComp-Plus-style fixed-corpus agent-search setting, where the submitted artifact is only the retriever checkpoint and the agent/judge/runtime are fixed externally.
Limitations
- The checkpoint is optimized for agentic retrieval from LRAT-style trajectory supervision.
- It should not be interpreted as a general-purpose factual QA model.
- Leaderboard results should be used only as checkpoint gates, not as sources for new training labels.