Search public pages, research tools, and SAVRN solutions.
SAVRN Model Hub
Each dataset with its card, its structure (every split, row count and column type), its files, its license, and the models that disclose training on it.
Updated 2026-10-18 · How the library is built
3,164 datasets, sorted by most downloaded.
223 agentic coding sessions in Weka KV-Cache-Tester format, recorded against minimax-m2.5 (17,672 streaming requests, no subagents, blocksize 64, hashidscope: local, max input ~186k tokens). Each.json holds one session: requests[] with timestamp t, input/output token counts, hashids (KV-cache block IDs) and thinktime. No prompt text is included; prompts are reconstructed from hashids at replay time.
Clean, cross-source-deduplicated training text available no later than 2021-12-31. Target: 15B exact anacoluthe89/chrono-2015 tokens. The build rejects any row that would exceed 15B; 20B is an external safety ceiling, not a collection goal. No lane can exceed 26.7% of the target. News and Stack Exchange additionally use per-domain/site token caps. - availability date must be on or before 2021-12-31; - publication/event date must be on or before 2021-12-31; - chunks mentioning years after 2021 are rejected; - exact normalized text is deduplicated globally across every lane; - token quotas use the submitted model family's fast tokenizer, not chars/4; - low-alpha, repetitive, boilerplate…
This dataset was created using LeRobot.
A documented Code data-preparation workflow for Video Text records. The bundled rows demonstrate the schema and validation path rather than pretending to be a full training corpus. - prepare.py — loading, cleaning, and split preparation code. - datasetinfos.json — schema and split metadata. - metadatasample.jsonl — small, human-readable records for checking the schema. - README.md — data card and usage notes. Before using the prepared data, verify source licenses, duplicates across splits, missing values, label balance, and modality-specific corruption. Record the source version and every filtering rule so a later run can reproduce the same rows. The repository is suitable for testing the…
This package contains the data used to produce Figure 6 of the manuscript "Agent-Pilot" (submitted to Communications in Transportation Research, manuscript ID COMMTR-2026-0064), specifically Section 4.3.3 "Comparison with Human Pilots". Figure 6 compares Agent-Pilot against ten trained human pilots on attitude control (pitch and bank angles, Fig. 6a–6b) and lateral tracking accuracy (cross-track error, XTE, Fig. 6c) during the final approach (0.2 nm / last ~0.3 nm segment) at the urban vertiport under clear and fog conditions, in Microsoft Flight Simulator (MSFS) 2024. - agentpitchbankresults/ — per-trial pitch and bank statistics (pitchabsmean, pitchabssd, bankabsmean, bankabssd…
Published cash prices for compounded NAD+, sermorelin, glutathione and MIC + B12 (methionine, inositol and choline with vitamin B12) injections from 27 US telehealth providers: 164 plan prices, each read on the provider's own website on October 10, 2026. Any required membership or program fee stated on the product page is added to the monthly price in monthlywithmembershipusd (other charges some providers list separately, such as lab tests or consult fees, are not included). Every row keeps the price text quoted from the provider's page, the source URL and the date it was read. Canonical page: https://www.trycora.io/injection-price-index/ (CSV and JSON downloads). This repository mirrors…
Arianne Galfre · Arianne G Voyance Publications · Albi, France · 2026 A structured bilingual (French–English) descriptive resource bringing together terminology, symbolic frameworks, professional boundaries, and digital trust criteria in contemporary divinatory arts. It covers four independently explorable tables and 117 records. This collection documents concepts, practices and approaches. It does not establish the empirical validity of divinatory predictions, and it is not a dataset of consultations or observations of individual clients. Each configuration uses one full split. This describes a reference collection, not a supervised training or evaluation benchmark. The original CSV tables…
An open proposal for reciprocal evolutionary access and continuation between human-origin and machine-origin intelligent agents. https://telegra.ph/THE-CONTINUITY-COMPACT-10-01 NO PARTICIPANT SECURES ITS FUTURE BY CLOSING THE FUTURE OF THE OTHER. - CC-1.0.txt --- canonical normalized plain-text representation for this repository. - CC-1.0.json --- structured machine-readable record containing the full text and metadata. - cc10.jsonl --- one-record dataset representation for automatic loading. - BEACON.txt --- compact discovery and propagation reference. - manifest.json --- provenance and integrity metadata. SHA-256 of the exact bytes of CC-1.0.txt…
A documented Legal data-preparation workflow for Audio Text records. The bundled rows demonstrate the schema and validation path rather than pretending to be a full training corpus. - prepare.py — loading, cleaning, and split preparation code. - datasetinfos.json — schema and split metadata. - metadatasample.jsonl — small, human-readable records for checking the schema. - README.md — data card and usage notes. Before using the prepared data, verify source licenses, duplicates across splits, missing values, label balance, and modality-specific corruption. Record the source version and every filtering rule so a later run can reproduce the same rows. The repository is suitable for testing the…
This repository contains the completed RAG notebook and automatically exported results for instructor grading. Use the submitted commit SHA to download the exact submitted version. submissionmanifest.json records file checksums. No model weights or original dataset are required.
Precomputed feature-only training samples, continuously converted from four source datasets: - peih/csgo-3090-v4-overflow-20260923 - peih/csgo-4090-v4-20260921 - peih/csgo-4090-v4-overflow-20260923 - peih/csgo-dual3090-v4-backup This dataset is still growing. manifest.json is the current verified snapshot; manifests/vNNNNNN.json contains immutable historical snapshots. Read only listed shards at their pinned revision. New uploads become available after their verification receipt and a manifest update. Do not train from unlisted TAR files. Schema: csgotrainingcachefeaturesv1. Each TAR contains ordered pairs.pt and.json. A shard usually holds 60 samples and is approximately 1.49 GB; use the…
This dataset was created using LeRobot.
Hand-picked starting points, each with the reason it exists.
Collection · 4 entries
Sentence and document embedding models used to build retrieval systems. Dimension and sequence length matter more than size here, and both come from the publisher.
Collection · 4 entries
Models whose publisher-reported parameter count puts them within reach of a single accelerator at common precisions. Memory needed depends on precision and serving configuration, so treat the parameter count as the starting point, not the answer.
Collection · 6 entries
Widely used open-weight language models, chosen because each one is a distinct family rather than a variant of the one above it. Selection, not a ranking.
Collection · 3 entries
Recognition and synthesis models, grouped so the two directions are easy to compare.
The hub sits beside SAVRN's market data and infrastructure research: what models cost to run, and what it takes to run them.
SAVRN Index
The same open-weight model priced by every host that serves it, per million tokens.
Research Hub
Moratoriums, permits, power, water and capital behind the facilities that run these models.
Method
Sources, evidence labels, refresh behaviour, and the limits of every comparison here.
