diff options
| author | YurenHao0426 <Blackhao0426@gmail.com> | 2026-02-16 14:44:42 -0600 |
|---|---|---|
| committer | YurenHao0426 <Blackhao0426@gmail.com> | 2026-02-16 14:44:42 -0600 |
| commit | 09d50e47860da0035e178a442dc936028808a0b3 (patch) | |
| tree | 9d651b0c7d289a9a0405953f2da989a3c431f147 /hag/pipeline.py | |
| parent | c90b48e3f8da9dd0f8d2ae82ddf977436bb0cfc3 (diff) | |
- Add centering support to MemoryBank (center_query, apply_centering, mean
persistence in save/load) to remove centroid attractor in Hopfield dynamics
- Add center flag to MemoryBankConfig, device field to PipelineConfig
- Grid search scripts: initial (β≤8), residual, high-β, and centered grids
with dedup-based LLM caching (89-91% call savings)
- Energy landscape visualization: 2D contour, 1D profile, UMAP, PCA heatmap
comparing centered vs uncentered dynamics
- Experiment log (note.md) documenting 4 rounds of results and root cause
analysis of centroid attractor problem
- Key finding: β_critical ≈ 37.6 for centered memory; best configs beat
FAISS baseline by +3-4% F1
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Diffstat (limited to 'hag/pipeline.py')
| -rw-r--r-- | hag/pipeline.py | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/hag/pipeline.py b/hag/pipeline.py index 1fefb84..086b3be 100644 --- a/hag/pipeline.py +++ b/hag/pipeline.py @@ -82,7 +82,8 @@ class RAGPipeline: if self.retriever_type == "hopfield": retrieval_result = self.hopfield_retriever.retrieve(query_emb) else: - query_np = query_emb.detach().numpy().astype(np.float32) + # FAISS requires CPU numpy arrays + query_np = query_emb.detach().cpu().numpy().astype(np.float32) retrieval_result = self.faiss_retriever.retrieve(query_np) # Generate |