Update model card with proper metadata and documentation

README.md

@@ -1,90 +1,76 @@
 ---
-language: en
+language:
+  - en
 license: mit
-library_name: transformers
-pipeline_tag: summarization
+library_name: onnx
 tags:
   - onnx
+  - summarization
+  - transformers.js
+  - in-browser
   - quantized
   - int8
-  - transformers.js
-  - summarization
   - bart
-  - browser-inference
+  - affectively
+  - edgework
+base_model: facebook/bart-large-cnn
+pipeline_tag: summarization
 datasets:
   - cnn_dailymail
-base_model: facebook/bart-large-cnn
 ---
 
-# BART-Large CNN Summarization (ONNX, Quantized int8)
+# BART Large CNN (ONNX, int8 Quantized)
 
-Quantized ONNX version of [facebook/bart-large-cnn](https://huggingface.co/facebook/bart-large-cnn) for efficient in-browser text summarization with [transformers.js](https://huggingface.co/docs/transformers.js).
+> **Production-ready** ONNX conversion of [facebook/bart-large-cnn](https://huggingface.co/facebook/bart-large-cnn) for in-browser text summarization — zero server cost, zero latency, complete privacy.
 
-## Overview
+## Highlights
 
-| | |
-|---|---|
-| **Task** | Abstractive text summarization |
-| **Base Model** | BART-large (406M params) |
-| **Format** | ONNX, dynamic quantization (int8) |
-| **Size** | ~493 MB (vs ~1.6 GB fp32) |
-| **Dataset** | [CNN/DailyMail](https://huggingface.co/datasets/cnn_dailymail) |
-| **Original Downloads** | 2.2M+ |
+- **Abstractive summarization** — generates concise summaries, not just extractive snippets
+- **~493 MB** quantized — BART-large architecture fine-tuned on CNN/DailyMail
+- **transformers.js compatible** — drop-in `pipeline('summarization')`
+- **406M parameters** — high-quality summaries from long-form text
 
-## Usage with transformers.js
+## Quick Start
 
 ```javascript
-import { pipeline } from '@xenova/transformers';
+import { pipeline } from '@huggingface/transformers';
 
 const summarizer = await pipeline(
   'summarization',
   'affectively-ai/bart-large-cnn-onnx',
-  { quantized: true }
+  { dtype: 'q8' }
 );
 
 const result = await summarizer(
-  'The tower is 324 metres (1,063 ft) tall, about the same height as an ' +
-  '81-storey building, and the tallest structure in Paris. Its base is ' +
-  'square, measuring 125 metres on each side. It was the first structure ' +
-  'to reach a height of 300 metres.',
-  { max_length: 60, min_length: 20 }
+  'Today was an incredibly challenging day at work. I had back-to-back meetings...',
+  { max_length: 50, min_length: 20 }
 );
-// [{ summary_text: 'The tower is 324 metres tall, about the same height as an 81-storey building. It was the first structure to reach a height of 300 metres.' }]
+// [{ summary_text: 'The author had a challenging day with back-to-back meetings...' }]
 ```
 
-## Why This Model?
+## Conversion Details
 
-BART-Large-CNN is the gold standard for English summarization:
-- **2.2M+ downloads** — the most popular summarization model on HuggingFace
-- Generates **fluent, abstractive summaries** (not just extractive)
-- Fine-tuned on **300K news articles** with human-written summaries
-- At **493 MB** quantized, it's practical for in-browser use
+| Property | Value |
+|----------|-------|
+| Base model | [facebook/bart-large-cnn](https://huggingface.co/facebook/bart-large-cnn) |
+| Training data | CNN/DailyMail (300k news articles) |
+| Export | PyTorch → ONNX via [Optimum](https://huggingface.co/docs/optimum) |
+| Quantization | int8 dynamic (`ORTQuantizer`, avx512_vnni) |
+| Quantized size | ~493 MB |
 
 ## Use Cases
 
-- Summarize journal entries in emotion tracking apps
-- Condense long therapy session notes
-- Create digest versions of articles or feedback
-- Summarize conversation threads
-
-## Conversion Details
-
-- **Exported** from PyTorch using [Optimum](https://huggingface.co/docs/optimum)
-- **Quantized** to int8 using `ORTQuantizer` with `avx512_vnni` config (encoder + decoder quantized separately)
-- **Converted by** [AFFECTIVELY](https://affectively.ai) via Cloud Build pipeline
-
-## Links
+This model powers text summarization in [Edgework.ai](https://edgework.ai) — bringing fast, cheap, and private inference as close to the user as possible. Best for:
 
-- **Website**: [aeonflux.dev](https://aeonflux.dev)
-- **Documentation**: [docs.aeonflux.dev](https://docs.aeonflux.dev)
-- **Storybook**: [storybook.aeonflux.dev](https://storybook.aeonflux.dev)
-- **GitHub**: [github.com/affectively-ai](https://github.com/affectively-ai)
-- **All Models**: [huggingface.co/affectively-ai](https://huggingface.co/affectively-ai)
+- Summarizing journal entries for daily/weekly emotion reports
+- Condensing therapy session notes
+- Creating digests from long conversation threads
+- Generating TL;DR for mental wellness content
 
-## About AFFECTIVELY
+## About
 
-[AFFECTIVELY](https://affectively.ai) is an emotion tracking and analysis platform. This model enables in-browser summarization of emotional journals and conversations.
+Published by [AFFECTIVELY](https://huggingface.co/affectively-ai) · Managed by [@buley](https://huggingface.co/buley)
 
-## License
+We convert, quantize, and publish **production-ready ONNX models** for edge and in-browser inference. Every release is tested for correctness and stability before publication.
 
-MIT (same as base model)
+- [All models](https://huggingface.co/affectively-ai) · [GitHub](https://github.com/affectively-ai) · [Edgework.ai](https://edgework.ai)