aeon-huggingface-trending · diff
git:20260512.574b491 to git:20260512.a4ba040
15 added, 69 removed. Audit A to A.
---
name: aeon-huggingface-trending
description: |
- Trending Hugging Face models, datasets, and spaces — filtered by relevance, clustered by theme,
- and labeled with a "why notable" line per pick. Goes beyond the raw download counter by surfacing
- what's actually shifting rather than just popular. Use for daily AI infra reads or to catch
- new model families before they mainstream.
+ Trending Hugging Face models, datasets, and spaces — filtered by license sanity, dedup vs same-week
+ quantizations, with a "why notable" line per pick (architecture shift, size step, license change,
+ notable author). Surfaces what's actually shifting rather than just popular.
Triggers: "trending on HF", "what models are hot", "huggingface trending", "new spaces today",
"best new datasets".
---
# aeon-huggingface-trending
- Daily filtered scan over Hugging Face's trending feeds. Three surfaces — models, datasets, spaces — each cluster-ranked rather than raw-download ranked.
+ Daily filtered scan over HF's three trending surfaces — models, datasets, spaces — cluster-ranked rather than raw-download ranked.
- ## Surfaces and HF endpoints
+ ## Endpoints
```bash
- # Trending models (last 24h)
curl -s "https://huggingface.co/api/models?sort=trending&direction=-1&limit=30"
-
- # Trending datasets
curl -s "https://huggingface.co/api/datasets?sort=trending&direction=-1&limit=30"
-
- # Trending spaces
curl -s "https://huggingface.co/api/spaces?sort=trending&direction=-1&limit=30"
-
- # Model detail (for the "why notable" line)
- curl -s "https://huggingface.co/api/models/${author}/${name}"
```
- ## Filtering rules
-
- A trending entry must survive:
+ ## Filters
- | Filter | Drops |
- |---|---|
- | **License sanity** | Models with no declared license or unclear commercial-use status get a flag, not a drop. |
- | **Repo size sanity** | Empty repos, no commits, or no model card → drop. |
- | **Quantization-only** | If the entry is a quantization of an already-trending model from the same week → demoted to a "quantizations" tail section. |
- | **Fork without delta** | Detected via README diff vs the upstream → drop. |
- | **Spam handles** | Authors with > 5 trending entries in 24h get demoted (typically aggregators). |
+ - Empty repos, no commits, no model card → drop.
+ - Same-week quantization of an already-trending model → demoted to "Quantizations" tail.
+ - Fork without README delta vs upstream → drop.
+ - Authors with > 5 trending entries in 24h → demoted (typically aggregators).
+ - License unclear or missing → flagged inline, not dropped.
## "Why notable" line
- Per surfaced entry, a one-sentence tag based on:
-
- - New architecture, novel training approach, first multi-modal in a domain, etc.
- - Significant size / context-window step (`128k`, `1M context`, `100B params`).
- - Author affiliation (major lab, prominent independent author).
- - License change (e.g. previously non-commercial → commercial).
-
- If no concrete reason exists, the entry says "no clear why — popular but unremarkable".
+ Per surfaced entry, a one-sentence tag: new architecture / size step / context-window jump / notable author affiliation / license change. If no concrete reason exists, the entry says "no clear why — popular but unremarkable" rather than inventing one.
## Output
- ```
- *HuggingFace Trending — 2026-05-12*
-
- Models (5)
- 1. lab/model-name (12k downloads / 24h)
- Why: first open-weight 1M-context model with full attention, not ring/sparse
- 2. author/finetune-name (8k)
- Why: domain-specific finetune of llama-4 on medical data, MIT license
- 3. team/release-v2 (6k)
- Why: replaces author's prior trending model with a method shift in §2 of the card
- ...
-
- Datasets (3)
- 1. lab/corpus-name (3k downloads)
- Why: synthetic dataset with verifiable provenance — first to attach generation logs
-
- Spaces (4)
- 1. author/demo-name (4k views)
- Why: live demo of yesterday's HF Papers pick; runs on a single A10
- 2. team/tool-name (2k)
- Why: agent-first tool — exposes itself via x402, not just web UI
-
- Quantizations (tail, 4 entries)
- - q/model-q4, q/model-q8, ... (low signal, listed for completeness)
- ```
+ Three sections — Models, Datasets, Spaces — each with the surviving picks, "why notable", license, and download/view count. Tail section for quantizations.
- ## Guidelines
+ ## Rules
- "Why notable" is a hard requirement. No reason → no surface.
- - Cluster quantizations away from headlines; they're useful but they're not the news.
- License flags appear inline (Apache 2.0, MIT, OpenRAIL-M, custom-no-commercial). Operators trading on model output care.
- - Spaces section often beats the models section for builder-tier signal — a working demo is a stronger validation than a card claim.
-
- ## Pairs with
-
- - `aeon-paper-pick` — yesterday's paper pick often appears as today's space.
- - `aeon-deep-research` when a trending entry merits a DD.
-
- ## Required keys
-
- None — Hugging Face API is public.
+ - Spaces section often beats Models for builder-tier signal — a working demo is stronger validation than a card claim.