Dataset Viewer
Auto-converted to Parquet Duplicate
asset_id
string
upload_id
string
title
string
external_id
string
creator_id
string
status
string
video_quality
string
resolution_tier
string
duration_seconds
float64
aspect_ratio
string
max_width
int64
max_height
int64
max_frame_rate
int64
encoding_tier
string
normalize_audio
bool
mp4_support
string
playback_policy
string
playback_id_public
string
video_track_id
string
audio_track_id
string
audio_primary
bool
audio_channels
int64
audio_language_code
string
created_at_unix
int64
raw_mux_json
dict
PLACEHOLDER_ASSET
PLACEHOLDER_UPLOAD
placeholder
0
0
ready
basic
720p
1
16:9
1,280
720
30
baseline
true
none
public
PLACEHOLDER_PLAYBACK
PLACEHOLDER_VIDEO_TRACK
PLACEHOLDER_AUDIO_TRACK
true
2
und
1,700,000,000
{ "placeholder": true }

MindsEye — Mux Assets Ledger Dataset

This repo is a ledger-style dataset for tracking Mux video assets as append-only records.

Instead of treating media as a one-off upload, we store every asset snapshot as an auditable entry:

  • playback IDs
  • track IDs (video/audio)
  • encoding + resolution metadata
  • status + progress
  • full raw payload for provenance

This aligns with the MindsEye principle:

Don’t overwrite intelligence. Record it.


Dataset Goal

Create a Hugging Face dataset where each row = one Mux asset snapshot.

You can later append new rows when:

  • audio is replaced
  • subtitles/transcripts are generated
  • new playback IDs appear
  • encoding tiers/resolutions change
  • policy changes occur

Folder Structure

mindseye-mux-assets-ledger/
  README.md
  data/
    mux_assets.jsonl
  scripts/
    build_dataset.py
    validate_schema.py
Downloads last month
29