ChicoPanama commited on
Commit
24ca49f
·
verified ·
1 Parent(s): eeeb7d6

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +87 -2
README.md CHANGED
@@ -1,3 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  # PokeDAO Mew-1A Training Dataset (Layered Architecture)
2
 
3
  🧬 **Project Mew-1A**: The world's first AI training dataset specifically for Pokemon TCG pricing analysis, extracted from a production-ready layered database architecture.
@@ -94,14 +162,29 @@ Each card is classified into **TCG Rarity-Based Price Tiers**:
94
  - **Task**: Text generation / Classification
95
  - **Fine-tuning**: LoRA (rank 8, alpha 16)
96
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
97
  ## Metadata
98
 
99
  - **Version**: v1
100
- - **Created**: 2025-10-06T01:15:40.013Z
101
  - **Source**: Layered Database (canonical_cards + market_listings)
102
  - **Tier System**: TCG Rarity-Based (Option B)
103
  - **Total Examples**: 10,000
104
- - **Duration**: 1.6s
105
 
106
  ## Data Sources
107
 
@@ -115,6 +198,8 @@ Each card is classified into **TCG Rarity-Based Price Tiers**:
115
 
116
  Proprietary - PokeDAO 2025
117
 
 
 
118
  ## Citation
119
 
120
  ```bibtex
 
1
+ ---
2
+ language:
3
+ - en
4
+ license: other
5
+ task_categories:
6
+ - text-classification
7
+ - text-generation
8
+ - tabular-classification
9
+ tags:
10
+ - pokemon
11
+ - tcg
12
+ - trading-cards
13
+ - pricing
14
+ - market-analysis
15
+ - collectibles
16
+ - finance
17
+ pretty_name: PokeDAO Mew-1A Training Dataset (Layered)
18
+ size_categories:
19
+ - 1K<n<10K
20
+ configs:
21
+ - config_name: default
22
+ data_files:
23
+ - split: train
24
+ path: data/train-*
25
+ dataset_info:
26
+ features:
27
+ - name: id
28
+ dtype: string
29
+ - name: card_name
30
+ dtype: string
31
+ - name: card_set
32
+ dtype: string
33
+ - name: card_number
34
+ dtype: string
35
+ - name: variant
36
+ dtype: string
37
+ - name: rarity
38
+ dtype: string
39
+ - name: is_first_edition
40
+ dtype: bool
41
+ - name: is_holo
42
+ dtype: bool
43
+ - name: is_reverse_holo
44
+ dtype: bool
45
+ - name: is_shadowless
46
+ dtype: bool
47
+ - name: avg_price_cents
48
+ dtype: int64
49
+ - name: avg_price_formatted
50
+ dtype: string
51
+ - name: total_listings
52
+ dtype: int64
53
+ - name: price_tier
54
+ dtype: string
55
+ - name: price_tier_emoji
56
+ dtype: string
57
+ - name: sample_listings
58
+ dtype: string
59
+ - name: data_quality
60
+ dtype: float64
61
+ splits:
62
+ - name: train
63
+ num_bytes: 5213440
64
+ num_examples: 10000
65
+ download_size: 2145280
66
+ dataset_size: 5213440
67
+ ---
68
+
69
  # PokeDAO Mew-1A Training Dataset (Layered Architecture)
70
 
71
  🧬 **Project Mew-1A**: The world's first AI training dataset specifically for Pokemon TCG pricing analysis, extracted from a production-ready layered database architecture.
 
162
  - **Task**: Text generation / Classification
163
  - **Fine-tuning**: LoRA (rank 8, alpha 16)
164
 
165
+ ## Training Configuration
166
+
167
+ ```yaml
168
+ model: meta-llama/Llama-3.2-3B-Instruct
169
+ task: text-generation
170
+ method: LoRA
171
+ parameters:
172
+ lora_rank: 8
173
+ lora_alpha: 16
174
+ learning_rate: 2e-4
175
+ epochs: 3
176
+ batch_size: 4
177
+ gradient_accumulation_steps: 4
178
+ ```
179
+
180
  ## Metadata
181
 
182
  - **Version**: v1
183
+ - **Created**: 2025-10-05
184
  - **Source**: Layered Database (canonical_cards + market_listings)
185
  - **Tier System**: TCG Rarity-Based (Option B)
186
  - **Total Examples**: 10,000
187
+ - **File Size**: ~5MB
188
 
189
  ## Data Sources
190
 
 
198
 
199
  Proprietary - PokeDAO 2025
200
 
201
+ For commercial use, please contact: [PokeDAO Team]
202
+
203
  ## Citation
204
 
205
  ```bibtex