datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
fixed-n-rb-er-cost-marginrl-qwen3-1.7b-base-math12k-token-mean-fixed-q0p8-run2-rollouts
fixed_n_rb_er_cost_marginrl_Qwen3-1.7B-Base_math12k_token_mean_fixed_q0.8_run2 rollouts
This dataset contains one compressed JSONL shard for every completed training
step. The step and rollout_index columns uniquely locate a rollout within
this training run. Run metadata and per-step row counts are recorded in
rollout_manifest.json.
fixed-n-rb-cost-aware-marginrl-qwen3-1.7b-base-math12k-token-mean-rerun-rollouts
fixed_n_rb_cost_aware_marginrl_Qwen3-1.7B-Base_math12k_token_mean_rerun rollouts
This dataset contains one compressed JSONL shard for every completed training
step. The step and rollout_index columns uniquely locate a rollout within
this training run. Run metadata and per-step row counts are recorded in
rollout_manifest.json.
fixed-n-rb-er-cost-marginrl-qwen3-1.7b-base-math12k-token-mean-run2-rollouts
fixed_n_rb_er_cost_marginrl_Qwen3-1.7B-Base_math12k_token_mean_run2 rollouts
This dataset contains one compressed JSONL shard for every completed training
step. The step and rollout_index columns uniquely locate a rollout within
this training run. Run metadata and per-step row counts are recorded in
rollout_manifest.json.
repro-memory-savings-at-what-cost-a-study-of-alternatives-to-backpropagation-traces
Agent traces
Agent sessions published from a Trackio Logbook.
agent-cost-traces
Agent Cost Traces: Synthetic Training Data
10,000 synthetic agent traces for training cost-aware model routers and agent optimizers.
Schema
Field
Type
Description
trace_id
string
Unique identifier
request
string
User request text
task_type
string
One of 9 task categories
difficulty
int
Estimated difficulty (1-5)
model_tier
int
Model tier used (1-5)
model_success
bool
Whether the model succeeded
optimal_tier
int
Minimum tier that would succeed… See the full description on the dataset page: https://huggingface.co/datasets/narcolepticchicken/agent-cost-traces.fixed-n-rb-offset-cost-aware-marginrl-qwen3-1.7b-base-math12k-offset2048-token-mean-rollouts
fixed_n_rb_offset_cost_aware_marginrl_Qwen3-1.7B-Base_math12k_offset2048_token_mean rollouts
This dataset contains one compressed JSONL shard for every completed training
step. The step and rollout_index columns uniquely locate a rollout within
this training run. Run metadata and per-step row counts are recorded in
rollout_manifest.json.
CostBench
CostBench
This dataset contains 381 records from CostBench_queries.json.
The queries are derived from the official CostBench benchmark repository and follow its travel-task query schema.
Contents
Top-level fields:
query_id, TimeInfo, task, is_location, goal_type, preferences, groundtruth, validation_raw, is_valid, user_requirements, query
Field Guide
query_id: Unique identifier for each query.
TimeInfo: Time context used in the task prompt. It is an ID-style… See the full description on the dataset page: https://huggingface.co/datasets/JiayuJeff/CostBench.er_cost_marginrl_r1_distill_1.5b_compression_n16_b512_32k_lr1e-6_kl0_seed42-rollouts
er_cost_marginrl_r1_distill_1.5b_compression_n16_b512_32k_lr1e-6_kl0_seed42 rollouts
This dataset contains one compressed JSONL shard for every completed training
step. The step and rollout_index columns uniquely locate a rollout within
this training run. Run metadata and per-step row counts are recorded in
rollout_manifest.json.
cost-analysis
Cost Analysis
Cloud API vs on-device inference cost comparison.
At 10K queries/day: Save $18,249/year with on-device.
At 100K queries/day: Save $182,499/year.
🚀 dispatchAI
namespace-cost-modelwikipedia-pt-br-domain
wikipedia-pt-br-domain-gemma
Versão enriquecida de costadev00/wikipedia-pt-br-extract com um label sintético de domínio por artigo.
Processo
Cada registro preserva os campos originais esperados da Wikipedia (page_id, title, text, ns, section_texts) e adiciona domain, derivado do label documental primary_category produzido pelo modelo.
Modelo
Modelo usado para labeling: google/gemma-4-26B-A4B-it.
Versão da pipeline: 0.1.0.
Limitações
O campo domain é… See the full description on the dataset page: https://huggingface.co/datasets/costadev00/wikipedia-pt-br-domain.low-high-cost-promptsnamespace-cost-modelwikipedia-pt-br-article-labels
wikipedia-pt-br-article-labels-gemma
Dataset sintético de labels documentais para artigos da Wikipedia em português brasileiro.
Origem
Os registros derivam de costadev00/wikipedia-pt-br-extract. Cada linha representa um artigo e preserva page_id, title, source_dataset e a licença herdada cc-by-sa-3.0.
Processo
A pipeline aplica triagem determinística em CPU, remove documentos ruins e usa um modelo Gemma local para gerar categorias, subcategorias, tipo… See the full description on the dataset page: https://huggingface.co/datasets/costadev00/wikipedia-pt-br-article-labels.wikipedia-pt-br-instructions
wikipedia-pt-br-instructions-gemma
Dataset sintético de Instruction Following em português brasileiro, derivado de artigos da Wikipedia pt-BR.
Origem
Os exemplos derivam de costadev00/wikipedia-pt-br-extract e preservam source_page_id, source_title, source_dataset e a licença herdada cc-by-sa-3.0.
Processo
A geração é inspirada em Alpaca e Self-Instruct: cada artigo válido passa por uma chamada de analista documental que produz candidatos de instrução ancorados… See the full description on the dataset page: https://huggingface.co/datasets/costadev00/wikipedia-pt-br-instructions.