concept
active
concept:qwen3-1-7bQwen3-1.7B
Smallest Qwen3 model tested; used in conscientiousness sweep example (Table 6)
Neighborhood — ranked by edge-count
Concepts (5)
concept
- Qwen3-4Brelated_to4B Qwen3 model tested in OCEAN benchmarks
- Qwen3.5-9Brelated_toSmallest model tested as evolver; produces harness updates comparable to Claude Opus 4.6 on SkillsBench
- Qwen3-32Brelated_toWeak-tier open-source model exhibiting both harness activation failure and adherence failure, with 25.1% skill-load rate
- Qwen3-14Brelated_to14B Qwen3 model quantized to 4-bit NF4; tested in OCEAN benchmarks
- Qwen2.5-VL-7Brelated_toBase vision-language model used to instantiate ATLAS.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Embedding model used to embed user messages for ridge regression analysis of persona drift causes
- Large open-source model used as anchor agent and anchor evolver; illustrates benchmark-dependent evolver performance
- One of four LLMs selected for representation analysis; embedding dimension D=4096; used as demonstration model in scatter plots.
- One of four LLMs selected; larger model with D=8192 embedding dimension; analyzed across proportionally aligned layers.
- Qwen 35B (3B active params, score 4.38) outscores Hermes 405B (405B active params, score 1.75) by 2.5xfinding0.738Parameters don't predict scores; 135x more parameters yields 60% lower score
- Quantifies harness activation failure for weak-tier models vs. strong-tier models
- Qwen3-235B leads as evolver on SWE-bench with 8.2 pp harness-updating gain but ranks last on MCP with 0.6 ppfinding0.725Illustrates benchmark-dependent reshuffling of evolver rankings, no evolver dominates across all substrates
- Strongest cross-family probe; explains clearer introspection in Qwen than Gemma