erenyanic commited on
Commit
b75df02
·
verified ·
1 Parent(s): c60ec61

Add YAML card metadata to fix empty repo-card metadata warning

Browse files
Files changed (1) hide show
  1. README.md +12 -0
README.md CHANGED
@@ -1,3 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
  # eldamar-tokenizer
2
 
3
  A tiny, end-to-end pipeline that **learns to invent Middle-earth names**. It cleans a dataset of Tolkien character and place names and trains four tiny from-scratch LLM architectures on it under three tokenisers — a plain character baseline plus character-level **BPE** at vocab 256 and 512 — for **12 models in total** (4 architectures × 3 tokenisers).
 
1
+ ---
2
+ language:
3
+ - en
4
+ tags:
5
+ - tokenizer
6
+ - bpe
7
+ - character-level
8
+ - text-generation
9
+ - tolkien
10
+ - middle-earth
11
+ ---
12
+
13
  # eldamar-tokenizer
14
 
15
  A tiny, end-to-end pipeline that **learns to invent Middle-earth names**. It cleans a dataset of Tolkien character and place names and trains four tiny from-scratch LLM architectures on it under three tokenisers — a plain character baseline plus character-level **BPE** at vocab 256 and 512 — for **12 models in total** (4 architectures × 3 tokenisers).