# Concepts

Second-brain semantic layer: nouns map atomically to file sets (EXTRACTED); verbs aggregate structural edges (INFERRED).

| Concept | Files | Mentions | Top Files |
|---------|-------|----------|-----------|
| `topo` | 27 | 99 | `app.py`, `eval/diag_static.py`, `eval/harness.py`, `eval/hodge_cm_ablation.py`, `eval/noise_sweep.py` |
| `model` | 25 | 186 | `eval/analyze_results.py`, `eval/governor.py`, `eval/governor_smoke.py`, `eval/harness.py`, `eval/noise_sweep.py` |
| `eval` | 25 | 77 | `eval/analyze.py`, `eval/analyze_results.py`, `eval/diag_static.py`, `eval/governor.py`, `eval/governor_smoke.py` |
| `run` | 22 | 70 | `app.py`, `eval/analyze_results.py`, `eval/governor.py`, `eval/harness.py`, `eval/integration_smoke.py` |
| `gpt3` | 20 | 48 | `app.py`, `eval/diag_static.py`, `eval/harness.py`, `eval/hodge_cm_ablation.py`, `eval/noise_sweep.py` |
| `topogpt3` | 20 | 38 | `app.py`, `eval/governor.py`, `eval/harness.py`, `topogpt3/__init__.py`, `topogpt3/__main__.py` |
| `load` | 19 | 40 | `eval/analyze.py`, `eval/analyze_results.py`, `eval/governor_smoke.py`, `eval/harness.py`, `eval/noise_analysis.py` |
| `build` | 18 | 44 | `app.py`, `eval/harness.py`, `eval/repair.py`, `eval/samplers.py`, `eval/sandbox.py` |
| `prompt` | 16 | 64 | `eval/analyze_results.py`, `eval/governor.py`, `eval/harness.py`, `eval/integration_smoke.py`, `eval/noise_analysis.py` |
| `returns` | 16 | 53 | `eval/governor.py`, `eval/harness.py`, `eval/samplers.py`, `eval/sandbox.py`, `synthetic_dataset.py` |
| `when` | 16 | 29 | `app.py`, `eval/analyze.py`, `eval/governor.py`, `eval/hodge_cm_ablation.py`, `eval/samplers.py` |
| `checkpoint` | 15 | 67 | `app.py`, `eval/diag_static.py`, `eval/hodge_cm_ablation.py`, `eval/noise_sweep.py`, `eval/smoke.py` |
| `config` | 15 | 58 | `eval/sandbox.py`, `eval/temp_sweep.py`, `tests/test_jlens.py`, `tests/test_lens_model.py`, `topogpt3/api_server.py` |
| `one` | 15 | 27 | `eval/analyze.py`, `eval/governor_smoke.py`, `eval/harness.py`, `eval/integration_smoke.py`, `eval/noise_sweep.py` |
| `all` | 14 | 37 | `eval/governor.py`, `eval/hodge_cm_ablation.py`, `eval/integration_smoke.py`, `eval/sandbox_smoke.py`, `tests/test_jlens.py` |
| `new` | 14 | 28 | `eval/governor.py`, `eval/harness.py`, `eval/integration_smoke.py`, `eval/repair.py`, `eval/samplers.py` |
| `runs` | 14 | 24 | `eval/analyze_results.py`, `eval/diag_static.py`, `eval/harness.py`, `eval/hodge_cm_ablation.py`, `eval/noise_analysis.py` |
| `layer` | 13 | 83 | `eval/hodge_cm_ablation.py`, `eval/sandbox.py`, `tests/test_jlens.py`, `tests/test_lens_model.py`, `topogpt3/api_server.py` |
| `per` | 13 | 52 | `eval/analyze.py`, `eval/governor.py`, `eval/governor_smoke.py`, `eval/noise_analysis.py`, `eval/sandbox.py` |
| `pass` | 13 | 37 | `eval/analyze.py`, `eval/harness.py`, `eval/integration_smoke.py`, `eval/noise_analysis.py`, `eval/noise_sweep.py` |
| `code` | 13 | 30 | `eval/governor.py`, `eval/harness.py`, `eval/samplers.py`, `eval/sandbox.py`, `run_hodge_cm_ablation.sh` |
| `token` | 12 | 88 | `eval/governor.py`, `eval/governor_smoke.py`, `eval/harness.py`, `eval/noise_analysis.py`, `infer_exploitgym.py` |
| `file` | 12 | 57 | `app.py`, `synthetic_dataset.py`, `tests/test_jlens.py`, `tests/test_lens_model.py`, `topogpt3/ewc.py` |
| `return` | 12 | 49 | `app.py`, `eval/governor.py`, `eval/harness.py`, `eval/sandbox.py`, `infer_exploitgym.py` |
| `error` | 12 | 34 | `eval/analyze.py`, `eval/harness.py`, `eval/report.py`, `eval/sandbox_smoke.py`, `tests/test_jlens.py` |
| `same` | 12 | 15 | `app.py`, `eval/governor.py`, `eval/harness.py`, `eval/repair.py`, `eval/sandbox.py` |
| `max` | 11 | 37 | `eval/diag_static.py`, `eval/governor.py`, `eval/sandbox.py`, `tests/test_lens_model.py`, `topogpt3/__init__.py` |
| `top` | 11 | 36 | `app.py`, `eval/governor.py`, `eval/harness.py`, `eval/repair.py`, `eval/sandbox.py` |
| `each` | 11 | 30 | `eval/analyze_results.py`, `eval/governor.py`, `eval/harness.py`, `eval/repair.py`, `eval/sandbox.py` |
| `train` | 11 | 28 | `run_exploitgym.sh`, `run_exploitgym_v2.sh`, `run_hodge_cm_ablation.sh`, `run_merged.sh`, `topogpt3/__init__.py` |
| `output` | 11 | 19 | `eval/harness.py`, `eval/repair.py`, `eval/samplers.py`, `eval/sandbox.py`, `eval/sandbox_smoke.py` |
| `usage` | 11 | 13 | `app.py`, `eval/analyze_results.py`, `eval/governor.py`, `eval/sandbox.py`, `run_exploitgym.sh` |
| `generation` | 10 | 35 | `eval/analyze.py`, `eval/governor.py`, `eval/governor_smoke.py`, `eval/harness.py`, `eval/noise_analysis.py` |
| `gpt2` | 10 | 31 | `eval/noise_sweep.py`, `synthetic_dataset.py`, `tests/test_lens_model.py`, `topogpt3/__init__.py`, `topogpt3/inference.py` |
| `text` | 10 | 31 | `app.py`, `eval/governor.py`, `eval/harness.py`, `infer_exploitgym.py`, `tests/test_lens_model.py` |
| `data` | 10 | 25 | `run_exploitgym.sh`, `run_merged.sh`, `synthetic_dataset.py`, `topogpt3/ewc.py`, `topogpt3/exploitgym_config.py` |
| `training` | 10 | 25 | `app.py`, `eval/harness.py`, `topogpt3/ewc.py`, `topogpt3/exploitgym_config.py`, `topogpt3/exploitgym_loader.py` |
| `weights` | 10 | 25 | `eval/hodge_cm_ablation.py`, `infer_exploitgym.py`, `run_exploitgym.sh`, `tests/test_lens_model.py`, `topogpt3/inference.py` |
| `single` | 10 | 24 | `eval/analyze.py`, `eval/governor.py`, `eval/harness.py`, `infer_exploitgym.py`, `synthetic_dataset.py` |
| `shape` | 10 | 23 | `eval/repair.py`, `tests/test_jlens.py`, `tests/test_lens_model.py`, `topogpt3/inference.py`, `topogpt3/inference_hrm.py` |
| `full` | 10 | 21 | `app.py`, `run_exploitgym.sh`, `synthetic_dataset.py`, `tests/test_lens_model.py`, `topogpt3/ewc.py` |
| `human` | 10 | 20 | `eval/analyze.py`, `eval/analyze_results.py`, `eval/harness.py`, `eval/integration_smoke.py`, `eval/noise_sweep.py` |
| `true` | 10 | 20 | `eval/governor.py`, `eval/integration_smoke.py`, `infer_exploitgym.py`, `synthetic_dataset.py`, `topogpt3/continuation.py` |
| `loader` | 10 | 19 | `eval/harness.py`, `eval/noise_sweep.py`, `eval/repair.py`, `topogpt3/exploitgym_config.py`, `topogpt3/exploitgym_loader.py` |
| `not` | 10 | 19 | `eval/governor.py`, `eval/harness.py`, `eval/samplers.py`, `eval/sandbox.py`, `tests/test_jlens.py` |
| `tensor` | 10 | 19 | `eval/governor.py`, `tests/test_lens_model.py`, `topogpt3/ewc.py`, `topogpt3/inference.py`, `topogpt3/inference_hrm.py` |
| `real` | 10 | 15 | `eval/diag_static.py`, `eval/governor.py`, `eval/harness.py`, `eval/hodge_cm_ablation.py`, `topogpt3/api_server.py` |
| `standard` | 10 | 14 | `app.py`, `eval/harness.py`, `eval/report.py`, `eval/samplers.py`, `eval/sandbox.py` |
| `python` | 10 | 13 | `eval/analyze_results.py`, `eval/governor.py`, `eval/harness.py`, `eval/noise_analysis.py`, `eval/sandbox.py` |
| `tokens` | 9 | 64 | `eval/governor.py`, `eval/noise_analysis.py`, `topogpt3/__init__.py`, `topogpt3/inference.py`, `topogpt3/inference_hrm.py` |

## Verb Edges

| Source | Verb | Target | Strength |
|--------|------|--------|----------|
| `topo` | `depends_on` | `topogpt3` | 1.00 |
| `topo` | `depends_on` | `model` | 0.96 |
| `gpt3` | `depends_on` | `topogpt3` | 0.85 |
| `gpt3` | `depends_on` | `model` | 0.81 |
| `gpt3` | `depends_on` | `topo` | 0.81 |
| `topo` | `depends_on` | `config` | 0.79 |
| `topo` | `depends_on` | `new` | 0.79 |
| `model` | `depends_on` | `topogpt3` | 0.77 |
| `topo` | `depends_on` | `checkpoint` | 0.75 |
| `topo` | `depends_on` | `returns` | 0.75 |
| `topo` | `depends_on` | `file` | 0.73 |
| `model` | `depends_on` | `topo` | 0.71 |
| `topo` | `depends_on` | `all` | 0.71 |
| `topo` | `depends_on` | `build` | 0.71 |
| `topo` | `depends_on` | `prompt` | 0.71 |
| `topo` | `depends_on` | `run` | 0.71 |
| `checkpoint` | `depends_on` | `topogpt3` | 0.69 |
| `gpt3` | `depends_on` | `new` | 0.69 |
| `topo` | `depends_on` | `when` | 0.69 |
| `checkpoint` | `depends_on` | `model` | 0.67 |
| `gpt3` | `depends_on` | `config` | 0.67 |
| `topo` | `depends_on` | `gpt2` | 0.67 |
| `topo` | `depends_on` | `layer` | 0.67 |
| `topo` | `depends_on` | `load` | 0.67 |
| `checkpoint` | `depends_on` | `topo` | 0.65 |
| `gpt3` | `depends_on` | `checkpoint` | 0.65 |
| `topo` | `depends_on` | `max` | 0.65 |
| `topo` | `depends_on` | `tensor` | 0.65 |
| `topogpt3` | `depends_on` | `model` | 0.65 |
| `gpt3` | `depends_on` | `returns` | 0.62 |
| `model` | `depends_on` | `returns` | 0.62 |
| `topo` | `depends_on` | `shape` | 0.62 |
| `topo` | `depends_on` | `token` | 0.62 |
| `topo` | `depends_on` | `tokens` | 0.62 |
| `topogpt3` | `depends_on` | `topo` | 0.62 |
| `gpt3` | `depends_on` | `prompt` | 0.60 |
| `model` | `depends_on` | `new` | 0.60 |
| `topo` | `depends_on` | `data` | 0.60 |
| `topo` | `depends_on` | `per` | 0.60 |
| `topo` | `depends_on` | `true` | 0.60 |
| `gpt3` | `depends_on` | `all` | 0.58 |
| `gpt3` | `depends_on` | `build` | 0.58 |
| `gpt3` | `depends_on` | `file` | 0.58 |
| `gpt3` | `depends_on` | `run` | 0.58 |
| `gpt3` | `depends_on` | `when` | 0.58 |
| `model` | `depends_on` | `config` | 0.58 |
| `topo` | `depends_on` | `gpt3` | 0.58 |
| `topo` | `depends_on` | `loader` | 0.58 |
| `topo` | `depends_on` | `training` | 0.58 |
| `gpt3` | `depends_on` | `layer` | 0.56 |

## Dialectic Prompts

- Thesis: `all` centralizes 14 files; Antithesis: `checkpoint` pulls 15 files with 9 shared (Jaccard 0.45); Synthesis: should they merge, split by layer, or keep `bridges` explicit?
- Thesis: `all` centralizes 14 files; Antithesis: `config` pulls 15 files with 9 shared (Jaccard 0.45); Synthesis: should they merge, split by layer, or keep `bridges` explicit?
- Thesis: `all` centralizes 14 files; Antithesis: `each` pulls 11 files with 6 shared (Jaccard 0.32); Synthesis: should they merge, split by layer, or keep `bridges` explicit?
- Thesis: `all` centralizes 14 files; Antithesis: `error` pulls 12 files with 7 shared (Jaccard 0.37); Synthesis: should they merge, split by layer, or keep `bridges` explicit?
- Thesis: `all` centralizes 14 files; Antithesis: `file` pulls 12 files with 10 shared (Jaccard 0.62); Synthesis: should they merge, split by layer, or keep `bridges` explicit?
- Thesis: `all` centralizes 14 files; Antithesis: `full` pulls 10 files with 6 shared (Jaccard 0.33); Synthesis: should they merge, split by layer, or keep `bridges` explicit?
- Thesis: `all` centralizes 14 files; Antithesis: `gpt2` pulls 10 files with 6 shared (Jaccard 0.33); Synthesis: should they merge, split by layer, or keep `bridges` explicit?
- Thesis: `all` centralizes 14 files; Antithesis: `gpt3` pulls 20 files with 9 shared (Jaccard 0.36); Synthesis: should they merge, split by layer, or keep `depends_on` explicit?
- Thesis: `all` centralizes 14 files; Antithesis: `layer` pulls 13 files with 9 shared (Jaccard 0.50); Synthesis: should they merge, split by layer, or keep `bridges` explicit?
- Thesis: `all` centralizes 14 files; Antithesis: `load` pulls 19 files with 8 shared (Jaccard 0.32); Synthesis: should they merge, split by layer, or keep `bridges` explicit?