Upload folder using huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
tags:
|
| 3 |
+
- from-scratch
|
| 4 |
+
- hierarchical-concept-diffusion
|
| 5 |
+
- synth
|
| 6 |
+
---
|
| 7 |
+
|
| 8 |
+
# ncp arm — Hierarchical Concept-Diffusion synth run
|
| 9 |
+
|
| 10 |
+
- arm: `ncp`
|
| 11 |
+
- params: 50,500,737
|
| 12 |
+
- steps: 12000
|
| 13 |
+
- seq_len: 256, batch: 32
|
| 14 |
+
- final val LM loss: **6.1264**
|
| 15 |
+
- data: PleIAs/SYNTH (en subset: query+reasoning+answer, streaming, gpt2 tokenizer)
|
| 16 |
+
- optimizer: AdamW lr 0.0001, wd 0.1, cosine, warmup 200
|
model.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7fd1046077019a951dc98f326e4e7a21ace193098d4096cdb8752f1a3b13f44f
|
| 3 |
+
size 279650571
|