replace card (plain text)

README.md

@@ -1,21 +1,11 @@
----
-library_name: transformers
-base_model: openai/gpt-oss-20b
-language: [en, ja]
-tags: []
-spec:
-  version: mini-v1
-  bake:
-    method: rank1_qkv_rmsnorm
-    gain: 0.5
-    concept_vec: concept_vec_v15k.pt
-    vector_shape: [24, 6, 2880]
-  data:
-    neutral_examples: 86376
-    persona_examples_used: 14388
-    files:
-      repo: narukijima/revolutioner
-      en: R_instruction_pairs_en.jsonl
-      ja: R_instruction_pairs_ja.jsonl
----
-Implementation and data are encoded in the YAML `spec` above.
+# revolutioner-mini-v1
+
+**Technical notes**
+
+- Base: `openai/gpt-oss-20b` (bf16)
+- Steering: rank-1 delta on Q/K/V across 24 layers (RMSNorm-aware)
+- Concept vector: `concept_vec_v15k.pt`, shape [24, 6, 2880], gain=0.5
+- Checkpoint: single baked weights (no LoRA/adapters; knowledge ≈ base)
+- Data used: neutral_examples=86376, pairs_used=14388
+- Source files: `narukijima/revolutioner` → `R_instruction_pairs_en.jsonl`, `R_instruction_pairs_ja.jsonl`
+- Inference: use base tokenizer & chat template