Datasetpublic

haxfenx/pokemon-llm-svg-bench-results

Pokémon LLM SVG Bench (V1.1) This is a snapshot of scored SVG drawings from the Pokémon LLM SVG Bench — LLMs try to draw Pokémon in SVG, and we score the results. Prefer reading the live site for context; for scoring rules and methodology, see About. Unofficial fan / research project. Pokémon images are copyrighted by The Pokémon Company and related companies involved in developing, operating, and managing the Pokémon series. Pokémon descriptions are sourced from… See the full description on the dataset page: https://huggingface.co/datasets/haxfenx/pokemon-llm-svg-bench-results.

sourceHugging Facemitupdated 2d agoView on Hugging Face

0likes234downloads

Dataset Card

Pokémon LLM SVG Bench (V1.1)

This is a snapshot of scored SVG drawings from the [Pokémon LLM SVG Bench](https://svg-bench.fenx.work/) — LLMs try to draw Pokémon in SVG, and we score the results. Prefer reading the live site for context; for scoring rules and methodology, see [About](https://svg-bench.fenx.work/about).

Unofficial fan / research project. Pokémon images are copyrighted by The Pokémon Company and related companies involved in developing, operating, and managing the Pokémon series. Pokémon descriptions are sourced from 「神奇宝贝百科」(wiki.52poke.com). SVGs here are model outputs, not official art.

What's in the box?

Roughly 56 published models × ~30 SVGs each ≈ 1680 rows (exact numbers depend on the Hub revision).

Models

—Arrow 1.1
—Claude Fable 5
—Claude Fable 5.1
—Claude Opus 4.7
—Claude Opus 4.8
—Claude Opus 5
—Claude Opus 5.5
—Claude Sonnet 4.6
—Composer 2
—Composer 2.5
—DeepSeek v4 Flash 0731
—DeepSeek v4 Flash Preview
—DeepSeek v4 Pro Preview
—DeepSeek-V4-Pro-0813
—Doubao-Seed-2.0-pro
—Doubao-Seed-2.1-pro
—Gemini 3 Flash
—Gemini 3.1 Flash-Lite
—Gemini 3.1 Pro
—Gemini 3.5 Flash
—Gemini 3.5 Flash-Lite
—Gemini 3.6 Flash
—Gemini 3.7 Flash
—Gemini 3.8 Flash
—GLM-5.1
—GLM-5.2
—GLM-5.3
—GLM-5.3-Flash
—GPT-5.5
—GPT-5.6 Sol
—GPT-5.6 Terra
—GPT-6 Astra
—GPT-6 Sol
—Grok 4.3
—Grok 4.5
—Grok 4.6
—Grok 4.7
—Hy3
—Hy3 preview
—Inkling
—Kimi K2.6
—Kimi K3
—Laguna S 2.1
—MiMo-V2.5
—MiMo-V2.5-Pro
—MiMo-V2.6-Flash
—MiMo-V2.6-Pro
—Muse Spark 1.2
—Muse Spark 1.3
—Qwen3.6-Max-Preview
—Qwen3.6-Plus
—Qwen3.7-Max
—Qwen3.7-Plus
—Qwen3.8-Max
—Step 3.5 Flash
—Step 3.7 Flash

Path	What
`benchmark.jsonl`	One row per submission (prompt, scores, model name, …) — `test` split
`svgs/{submission_id}.svg`	The SVG file (`svg_file_name` points here)

Same model + same Pokémon often appears twice (two independent draws). Use submission_id to tell them apart.

In Dataset Viewer, svg_file_name is just a path — open Files → `svgs/` to peek at the drawings.

How it works (tiny version)

1.Prompt like: Generate the SVG code for the Pokémon Pikachu. Some models get an extra “SVG only” constraint; full text is in prompt.
2.We keep the SVG.
3.V1.1 scores three dimensions + a weighted score (0–10), by stage S1 / S2 / S3.

Details live on the About page — this Card stays light on purpose.

Fields (cheat sheet)

Field	Notes
`submission_id`	Unique drawing id
`svg_file_name`	e.g. `svgs/<id>.svg`
`prompt`	What the model was asked
`pokemon_name` / `stage`	Species + `S1`–`S3`
`model_display_name` / `model_notes`	Leaderboard name + optional notes
`leaderboard_published_at`	When the model went public on V1.1
`dimension_1_score` … `_3` / `weighted_score`	Scores 0–10
`dimension_*_label_en`	Dimension names for that stage
`score_version`	Always `v1.1` here

Also may include source_type and light SVG stats (tag_count, byte_length, advanced_usage_pct).

Dimension keys

—S1: geometry_precision, color_fidelity, overall_impression
—S2: signature_features, spatial_proportion, overall_impression
—S3: structural_decomposition, detail_polish, overall_impression

License

MIT for the dataset packaging and metadata in this repo. Pokémon IP and third-party model terms still apply to names / generations — don't pass these off as official art.