Hugging Face's logo

Datasets:
brianward92
/
draftfm-frozen-eval

Modalities:
Tabular
Text
Formats:
parquet
Size:
10K - 100K
Tags:
magic-the-gathering
draft
model-evaluation
reproducibility
predictions
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
Dataset Viewer
Auto-converted to Parquet Duplicate
draft_id
stringlengths
32
32
pack_number
int64
0
2
pick_number
int64
0
13
pack_size
int64
1
14
wr_bucket
float64
0.56
0.72
n_games_bucket
int64
100
1k
target_rank
int64
1
13
pick_prob
float64
0
1
top_prob
float64
0.16
1
7160ce977cc2476c8f226dce2474b362
0
0
14
0.56
100
1
0.489816
0.489816
7160ce977cc2476c8f226dce2474b362
0
1
13
0.56
100
1
0.43679
0.43679
7160ce977cc2476c8f226dce2474b362
0
2
12
0.56
100
2
0.149861
0.482214
7160ce977cc2476c8f226dce2474b362
0
3
11
0.56
100
3
0.208779
0.45705
7160ce977cc2476c8f226dce2474b362
0
4
10
0.56
100
1
0.41825
0.41825
7160ce977cc2476c8f226dce2474b362
0
5
9
0.56
100
1
0.935867
0.935867
7160ce977cc2476c8f226dce2474b362
0
6
8
0.56
100
1
0.870793
0.870793
7160ce977cc2476c8f226dce2474b362
0
7
7
0.56
100
3
0.010713
0.955855
7160ce977cc2476c8f226dce2474b362
0
8
6
0.56
100
1
0.685122
0.685122
7160ce977cc2476c8f226dce2474b362
0
9
5
0.56
100
2
0.199515
0.547445
7160ce977cc2476c8f226dce2474b362
0
10
4
0.56
100
1
0.484173
0.484173
7160ce977cc2476c8f226dce2474b362
0
11
3
0.56
100
2
0.275798
0.72355
7160ce977cc2476c8f226dce2474b362
0
12
2
0.56
100
1
0.750198
0.750198
7160ce977cc2476c8f226dce2474b362
0
13
1
0.56
100
1
1
1
7160ce977cc2476c8f226dce2474b362
1
0
14
0.56
100
1
0.932365
0.932365
7160ce977cc2476c8f226dce2474b362
1
1
13
0.56
100
2
0.009503
0.988472
7160ce977cc2476c8f226dce2474b362
1
2
12
0.56
100
7
0.012898
0.467385
7160ce977cc2476c8f226dce2474b362
1
3
11
0.56
100
1
0.687566
0.687566
7160ce977cc2476c8f226dce2474b362
1
4
10
0.56
100
2
0.390861
0.450958
7160ce977cc2476c8f226dce2474b362
1
5
9
0.56
100
3
0.03302
0.572611
7160ce977cc2476c8f226dce2474b362
1
6
8
0.56
100
1
0.774995
0.774995
7160ce977cc2476c8f226dce2474b362
1
7
7
0.56
100
1
0.87911
0.87911
7160ce977cc2476c8f226dce2474b362
1
8
6
0.56
100
1
0.898005
0.898005
7160ce977cc2476c8f226dce2474b362
1
9
5
0.56
100
1
0.994822
0.994822
7160ce977cc2476c8f226dce2474b362
1
10
4
0.56
100
1
0.564333
0.564333
7160ce977cc2476c8f226dce2474b362
1
11
3
0.56
100
1
0.624625
0.624625
7160ce977cc2476c8f226dce2474b362
1
12
2
0.56
100
1
0.999504
0.999504
7160ce977cc2476c8f226dce2474b362
1
13
1
0.56
100
1
1
1
7160ce977cc2476c8f226dce2474b362
2
0
14
0.56
100
1
0.887544
0.887544
7160ce977cc2476c8f226dce2474b362
2
1
13
0.56
100
2
0.387768
0.461468
7160ce977cc2476c8f226dce2474b362
2
2
12
0.56
100
3
0.023646
0.915083
7160ce977cc2476c8f226dce2474b362
2
3
11
0.56
100
2
0.237932
0.628596
7160ce977cc2476c8f226dce2474b362
2
4
10
0.56
100
1
0.415474
0.415474
7160ce977cc2476c8f226dce2474b362
2
5
9
0.56
100
2
0.213375
0.69288
7160ce977cc2476c8f226dce2474b362
2
6
8
0.56
100
1
0.825747
0.825747
7160ce977cc2476c8f226dce2474b362
2
7
7
0.56
100
3
0.185836
0.348821
7160ce977cc2476c8f226dce2474b362
2
8
6
0.56
100
1
0.742989
0.742989
7160ce977cc2476c8f226dce2474b362
2
9
5
0.56
100
4
0.144298
0.436966
7160ce977cc2476c8f226dce2474b362
2
10
4
0.56
100
1
0.89977
0.89977
7160ce977cc2476c8f226dce2474b362
2
11
3
0.56
100
1
0.619377
0.619377
7160ce977cc2476c8f226dce2474b362
2
12
2
0.56
100
2
0.268405
0.731595
7160ce977cc2476c8f226dce2474b362
2
13
1
0.56
100
1
1
1
8fe8ab8698b546619a73129413aca7f8
0
0
14
0.64
100
1
0.998345
0.998345
8fe8ab8698b546619a73129413aca7f8
0
1
13
0.64
100
2
0.098374
0.854394
8fe8ab8698b546619a73129413aca7f8
0
2
12
0.64
100
3
0.1469
0.375202
8fe8ab8698b546619a73129413aca7f8
0
3
11
0.64
100
2
0.342362
0.421087
8fe8ab8698b546619a73129413aca7f8
0
4
10
0.64
100
1
0.618724
0.618724
8fe8ab8698b546619a73129413aca7f8
0
5
9
0.64
100
1
0.967725
0.967725
8fe8ab8698b546619a73129413aca7f8
0
6
8
0.64
100
1
0.669901
0.669901
8fe8ab8698b546619a73129413aca7f8
0
7
7
0.64
100
2
0.22769
0.699347
8fe8ab8698b546619a73129413aca7f8
0
8
6
0.64
100
1
0.917728
0.917728
8fe8ab8698b546619a73129413aca7f8
0
9
5
0.64
100
3
0.219499
0.366902
8fe8ab8698b546619a73129413aca7f8
0
10
4
0.64
100
1
0.927911
0.927911
8fe8ab8698b546619a73129413aca7f8
0
11
3
0.64
100
2
0.065285
0.934096
8fe8ab8698b546619a73129413aca7f8
0
12
2
0.64
100
2
0.416255
0.583745
8fe8ab8698b546619a73129413aca7f8
0
13
1
0.64
100
1
1
1
8fe8ab8698b546619a73129413aca7f8
1
0
14
0.64
100
1
0.434294
0.434294
8fe8ab8698b546619a73129413aca7f8
1
1
13
0.64
100
2
0.146049
0.691836
8fe8ab8698b546619a73129413aca7f8
1
2
12
0.64
100
1
0.971524
0.971524
8fe8ab8698b546619a73129413aca7f8
1
3
11
0.64
100
1
0.571827
0.571827
8fe8ab8698b546619a73129413aca7f8
1
4
10
0.64
100
1
0.731213
0.731213
8fe8ab8698b546619a73129413aca7f8
1
5
9
0.64
100
1
0.566628
0.566628
8fe8ab8698b546619a73129413aca7f8
1
6
8
0.64
100
1
0.446602
0.446602
8fe8ab8698b546619a73129413aca7f8
1
7
7
0.64
100
1
0.682563
0.682563
8fe8ab8698b546619a73129413aca7f8
1
8
6
0.64
100
1
0.941021
0.941021
8fe8ab8698b546619a73129413aca7f8
1
9
5
0.64
100
1
0.96771
0.96771
8fe8ab8698b546619a73129413aca7f8
1
10
4
0.64
100
1
0.997062
0.997062
8fe8ab8698b546619a73129413aca7f8
1
11
3
0.64
100
1
0.814351
0.814351
8fe8ab8698b546619a73129413aca7f8
1
12
2
0.64
100
1
0.993254
0.993254
8fe8ab8698b546619a73129413aca7f8
1
13
1
0.64
100
1
1
1
8fe8ab8698b546619a73129413aca7f8
2
0
14
0.64
100
3
0.111688
0.431261
8fe8ab8698b546619a73129413aca7f8
2
1
13
0.64
100
1
0.780543
0.780543
8fe8ab8698b546619a73129413aca7f8
2
2
12
0.64
100
2
0.262868
0.379421
8fe8ab8698b546619a73129413aca7f8
2
3
11
0.64
100
2
0.191657
0.529764
8fe8ab8698b546619a73129413aca7f8
2
4
10
0.64
100
1
0.792147
0.792147
8fe8ab8698b546619a73129413aca7f8
2
5
9
0.64
100
1
0.733239
0.733239
8fe8ab8698b546619a73129413aca7f8
2
6
8
0.64
100
2
0.241619
0.610675
8fe8ab8698b546619a73129413aca7f8
2
7
7
0.64
100
2
0.374082
0.571011
8fe8ab8698b546619a73129413aca7f8
2
8
6
0.64
100
3
0.153133
0.399353
8fe8ab8698b546619a73129413aca7f8
2
9
5
0.64
100
1
0.77491
0.77491
8fe8ab8698b546619a73129413aca7f8
2
10
4
0.64
100
1
0.55724
0.55724
8fe8ab8698b546619a73129413aca7f8
2
11
3
0.64
100
2
0.428497
0.570431
8fe8ab8698b546619a73129413aca7f8
2
12
2
0.64
100
2
0.467236
0.532764
8fe8ab8698b546619a73129413aca7f8
2
13
1
0.64
100
1
1
1
e8123f6a615a4c7b9799db36899aeafb
0
0
14
0.58
100
1
0.924973
0.924973
e8123f6a615a4c7b9799db36899aeafb
0
1
13
0.58
100
1
0.349998
0.349998
e8123f6a615a4c7b9799db36899aeafb
0
2
12
0.58
100
2
0.035743
0.909391
e8123f6a615a4c7b9799db36899aeafb
0
3
11
0.58
100
1
0.77624
0.77624
e8123f6a615a4c7b9799db36899aeafb
0
4
10
0.58
100
1
0.913609
0.913609
e8123f6a615a4c7b9799db36899aeafb
0
5
9
0.58
100
2
0.123715
0.673342
e8123f6a615a4c7b9799db36899aeafb
0
6
8
0.58
100
3
0.261898
0.374291
e8123f6a615a4c7b9799db36899aeafb
0
7
7
0.58
100
1
0.487968
0.487968
e8123f6a615a4c7b9799db36899aeafb
0
8
6
0.58
100
2
0.129119
0.792245
e8123f6a615a4c7b9799db36899aeafb
0
9
5
0.58
100
1
0.833252
0.833252
e8123f6a615a4c7b9799db36899aeafb
0
10
4
0.58
100
2
0.103853
0.879634
e8123f6a615a4c7b9799db36899aeafb
0
11
3
0.58
100
1
0.532567
0.532567
e8123f6a615a4c7b9799db36899aeafb
0
12
2
0.58
100
1
0.988384
0.988384
e8123f6a615a4c7b9799db36899aeafb
0
13
1
0.58
100
1
1
1
e8123f6a615a4c7b9799db36899aeafb
1
0
14
0.58
100
1
0.414692
0.414692
e8123f6a615a4c7b9799db36899aeafb
1
1
13
0.58
100
4
0.083813
0.488833
End of preview. Expand in Data Studio

DraftFM frozen_eval: the per-pick prediction archive

Every prediction behind the tables in the DraftFM paper, archived so the results can be recomputed (no GPU needed) and audited pick by pick.

Each top-level directory is named by the sha256 of the frozen data snapshot it was evaluated against — the directory name itself is the registration. Inside are per-pick prediction parquets for every model in the battery (scaling rungs s1–s16, f-dev, f-full, ablations, and baselines), in both human and deployment scoring modes, plus derived accuracy curves and a summary.json recording the evaluation context and gates.

Directories

Directory What it is
013df16b…183a The headline MSH pass: single pre-registered zero-shot evaluation over the first public MSH snapshot (PremierDraft, protocol v1.1, executed 2026-07-07). Source of the 57.0% top-1 result
013df16b…183a_postday1 Post-day-one MSH ceiling: per-set model fine-tuned on MSH after the zero-shot pass, as pre-registered in the protocol (2026-07-09)
df6220d8…853 Full-battery end-to-end rehearsal on SOS PremierDraft (protocol v1, 2026-07-05), run before the real MSH pass to burn in the pipeline
6552b298…34e Earlier SOS TradDraft rehearsal with baselines (protocol v1, 2026-07-04)
validation/ Per-set validation parquets (e.g. SOS) used for sanity gates

The rehearsals are included deliberately: the protocol's credibility comes from showing the pipeline was exercised and frozen before the one-shot MSH evaluation, not tuned after it.

Using it

Each <model>.<mode>.parquet holds one row per pick with the model's probability over the pack and the chosen card; summary.json in each directory records the snapshot sha256, S3 ETag, protocol tag, execution timestamp, and data-quality gates. To recompute a table, aggregate the parquets; to audit a claim, join any pick back to its prediction. The battery definition lives in the model repo's frozen_battery.json; the chronology is in the code repo's docs/eval_protocol.md.

Provenance and licensing

These files are model outputs and derived aggregates generated by the DraftFM pipeline (Apache-2.0). The underlying human draft data is the 17Lands public dataset (CC BY 4.0 — "Data from 17Lands.com"); raw 17Lands rows are not redistributed here. Unofficial Fan Content per the Wizards of the Coast Fan Content Policy; not approved or endorsed by Wizards.

Contact

Brian Ward — brian.ward.92@gmail.com
Downloads last month
21
Number of rows:
69,803
Total file size:
1.75 GB