Simplify fixed proposals and remove cached combination scores

README.md

@@ -17,10 +17,9 @@ The matching checkpoints are in [alien-models](https://huggingface.co/alejandroh
 - `data.tar.gz`: SQLite paper and author metadata, original token vocabulary, and coherence train/validation splits. Extracts to `data/`.
 - `availability-compact.tar.gz`: the corrected availability dataset used by the released dual encoder. Extracts to `data/availability_dataset/`; approximately 1.8 GB compressed and 4.3 GB extracted.
 - `contrastive_availability_train.jsonl.gz` and `contrastive_availability_val.jsonl.gz`: optional exchange-format examples, approximately 3.75 GB and 0.42 GB compressed. The metadata and token mapping are provided alongside them.
-- `triplet_scores.npz`: optional cached coherence and availability scores for all 3,353,896 triplets over 273 atoms. Arrays are `combos`, `coherence`, and `availability`.
-- `research-assets.tar.gz`: optional original selection metadata and available decoded proposals, kept separate from the updated numerical top-300 selection. See its README for coverage.
+- `research-assets.tar.gz`: fixed atom combinations and Markdown proposal texts from Alien (299), Random (300), Claude Opus 4.7 (300), and Gemini 3.1 Pro (300). Each method folder contains a `combinations.json` index linking atom IDs to proposal files.
 
-The code README provides extraction and training commands. The compact representation is recommended for retraining.
+Download only the files needed for your workflow. The code README provides extraction and training commands. The compact representation is recommended for retraining.
 
 ```bash
 hf download alejandrohdez/alien-data papers.tar.gz data.tar.gz --repo-type dataset --local-dir downloads

manifest.json

@@ -1,8 +1,8 @@
 {
   "files": {
     "README.md": {
-      "bytes": 4163,
-      "sha256": "7db9f1c52e39c6be6a4894efde5c3d60b25cd9ed1e71ffcc0537447def5c4eec"
+      "bytes": 3835,
+      "sha256": "8aef157aafd3678fb1e203c7b54b30f18b2336a27ca3d9d9d90e97bae1b66b2e"
     },
     "availability-compact.tar.gz": {
       "bytes": 1806072602,
@@ -33,12 +33,8 @@
       "sha256": "a7699f3aa3ec9379d321f9ffc65b8bcf18dcb5895e9a61f4d2abad5a520aa10d"
     },
     "research-assets.tar.gz": {
-      "bytes": 4524782,
-      "sha256": "573c91efa10e36479f8d3bf61cf446923521b2e5d0e8085b4ef0afd5f08fb7e9"
-    },
-    "triplet_scores.npz": {
-      "bytes": 31461977,
-      "sha256": "33326e7d5970e58526e6bffcafa34ed27e4b91d39b80d41e1ff8238731c40376"
+      "bytes": 3053006,
+      "sha256": "258cc8802ffb9a662de566ca5c88f3b07af91d309e7245c4028f75603a3e1ff1"
     },
     "LICENSE": {
       "bytes": 1085,

research-assets.tar.gz

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:573c91efa10e36479f8d3bf61cf446923521b2e5d0e8085b4ef0afd5f08fb7e9
-size 4524782
+oid sha256:258cc8802ffb9a662de566ca5c88f3b07af91d309e7245c4028f75603a3e1ff1
+size 3053006

triplet_scores.npz

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:33326e7d5970e58526e6bffcafa34ed27e4b91d39b80d41e1ff8238731c40376
-size 31461977