Fix README front matter formatting
Browse files
README.md
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
-
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
- lora
|
| 8 |
- transformers
|
| 9 |
- cot-oracle
|
|
@@ -12,13 +12,13 @@
|
|
| 12 |
- ours
|
| 13 |
- 1-layer
|
| 14 |
- 22.5m-train-tokens
|
| 15 |
-
|
| 16 |
|
| 17 |
-
|
| 18 |
|
| 19 |
-
|
| 20 |
|
| 21 |
-
|
| 22 |
|
| 23 |
- Base model: `Qwen/Qwen3-8B`
|
| 24 |
- Adapter format: PEFT LoRA
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: Qwen/Qwen3-8B
|
| 3 |
+
library_name: peft
|
| 4 |
+
pipeline_tag: text-generation
|
| 5 |
+
tags:
|
| 6 |
+
- base_model:adapter:Qwen/Qwen3-8B
|
| 7 |
- lora
|
| 8 |
- transformers
|
| 9 |
- cot-oracle
|
|
|
|
| 12 |
- ours
|
| 13 |
- 1-layer
|
| 14 |
- 22.5m-train-tokens
|
| 15 |
+
---
|
| 16 |
|
| 17 |
+
# CoT Oracle Paper Ablation: Ours, 1 Layer
|
| 18 |
|
| 19 |
+
This repo contains the 1-layer paper ablation for the CoT Oracle recipe: on-policy lens tasks, chunked ConvQA, FineWeb lens readouts, and classification, without LatentQA.
|
| 20 |
|
| 21 |
+
## What This Checkpoint Is
|
| 22 |
|
| 23 |
- Base model: `Qwen/Qwen3-8B`
|
| 24 |
- Adapter format: PEFT LoRA
|