odyn
Datasets
All datasets matching “odyn”odyn-sharegpt-benchmarks
Odyn ShareGPT Benchmarks
This dataset repository contains fixed-size ShareGPT benchmark samples used for stress and orchestration testing.
Files
data/sharegpt_1000.jsonl (1,000 rows)
data/sharegpt_5000.jsonl (5,000 rows)
data/sharegpt_10000.jsonl (10,000 rows)
Source
Derived from RyokoAI/ShareGPT52K train split via deterministic sampling (seed = 7).
benchmark-dataset-different-gpu-workload
GPU catalog × LLM workload VRAM benchmark
Summary
Tabular benchmark in CSV form: each row pairs a catalog GPU (gpu_id, gpu_display_name, catalog_gpu_vram_gb) with a concrete LLM inference-style workload (model, parameter count, context length, precision, batch size, concurrent users). The file records math_engine VRAM component estimates (weights, KV cache, activations, overhead, totals, tier), a document_engine recommended VRAM value, a short comparison summary… See the full description on the dataset page: https://huggingface.co/datasets/odyn-network/benchmark-dataset-different-gpu-workload.finetune-tracker-v1merlinodyn-benchmarks
Odyn Benchmarks
Inference benchmark datasets and results for the Odyn Network — a distributed, OpenAI-compatible AI inference platform built on vLLM, Ray Serve, and FastAPI.
Dataset Structure
Prompt Profiles (data/)
Four load profiles covering the full input/output token distribution space, sourced from real Odyn traffic and augmented with ShareGPT Vicuna Unfiltered:
Profile
Description
Input tokens
Output tokens
Rows
A
Short input, Long output
avg… See the full description on the dataset page: https://huggingface.co/datasets/odyn-network/odyn-benchmarks.benchmark-finetune-dpo-v1
Odyn benchmark: DPO LoRA fine-tuning peak VRAM (V1)
Curated benchmark rows for validating GPU memory estimators during DPO + LoRA fine-tuning. Each row pairs a published or measured expected peak VRAM with inputs to a math engine (model size, context length, batch, LoRA rank, precision, parallelism) plus optional VRAM breakdown and provenance.
This dataset is not preference-pair training JSONL (UltraFeedback-style). It is evaluation ground truth for placement / scheduler memory… See the full description on the dataset page: https://huggingface.co/datasets/odyn-network/benchmark-finetune-dpo-v1.