Geocities Prompt Html: Fine-Tuned Gemma Model for Vintage HTML Generation

A dataset for fine-tuning the Gemma-4-E2B-it model to generate vintage-style HTML pages from plain-language prompts. The repository includes training data in JSONL format and associated scripts. The dataset was created by thekevinscott and last updated on 2026-06-19.

Use Cases

• Fine-tune language models for text-to-HTML generation based on the provided prompt-html pairs.
• Generate full, vintage-style HTML pages from natural language descriptions.
• Study the performance of LoRA fine-tuning on a specific code generation task.

Strengths

• Includes training scripts configured for a 32 GB GPU, providing a ready-to-run environment.
• Dataset format is explicitly defined as one JSONL object per line with 'prompt' and 'html' keys.

Limitations

• Dataset size, row count, and column details are unknown, limiting suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• License information is unavailable, which may restrict usage.

Provenance

Source

huggingface

Collection Method

Likely created for model fine-tuning, but specific gathering method is not described.

Freshness

Last updated 2026-06-19 02:39:44.