id stringlengths 20 20 | seed_id stringclasses 90
values | domain stringclasses 6
values | difficulty_elo int64 3.4k 3.4k | format stringclasses 1
value | instruction stringclasses 90
values | completion stringlengths 1.14k 58k |
|---|---|---|---|---|---|---|
UNIV-AGNT-000002-R06 | UNIV-AGNT-000002 | system_administration | 3,400 | universal_sft_rlef | A production Ubuntu 22.04 LTS server running Nginx 1.24, PostgreSQL 15, Redis 7, and a Go-based API service behind HAProxy is exhibiting intermittent connection timeouts, high iowait, and sporadic OOM kills during traffic spikes above 800 req/s. The system has LVM on mdadm RAID-10 with 4x NVMe SSDs, 128GB RAM, and cust... | <thought>
Deconstruct: Deliver exact diagnostic sequence using ONLY listed live tools (ss, iostat, iotop, perf, bpftrace, journalctl, strace on postgres/haproxy, pg_stat_statements). Must include commands, plausible expected outputs, and root-cause hypotheses tied to symptoms (timeouts, high iowait, OOM at >800 rps). T... |
UNIV-AGNT-000001-R06 | UNIV-AGNT-000001 | software_engineering | 3,400 | universal_sft_rlef | You are debugging a production Go microservice that processes financial trades using PostgreSQL with READ COMMITTED isolation. Under sustained 50k TPS load with 200ms network jitter to replicas, duplicate trade executions and lost updates occur during failover. Perform step-by-step root-cause analysis covering transact... | <thought>
Deconstruct objectives: Identify root causes of duplicate trade executions and lost updates in a Go PostgreSQL microservice under 50k TPS with READ COMMITTED, 200ms replica jitter, and failover; then produce production-ready fixes via idempotency, 2PC emulation, schema, observability, and validation harness. ... |
UNIV-AGNT-000003-R01 | UNIV-AGNT-000003 | data_engineering | 3,400 | universal_sft_rlef | Design a highly scalable real-time data pipeline capable of processing 500,000 events per second from IoT sensors worldwide. Utilize Apache Kafka for ingestion, Apache Flink for stream processing with complex event processing for anomaly detection, and AWS Redshift for the serving layer. Incorporate data quality checks... | <thought>
Deconstruct objectives: Deliver a complete, production-grade pipeline spec for 500k events/sec IoT ingestion/processing/serving with exact required components (Kafka, Flink CEP + stateful, Redshift), plus data quality, GDPR anonymization, autoscaling, Schema Registry, diagram, justifications/alternatives, Jav... |
UNIV-AGNT-000002-R01 | UNIV-AGNT-000002 | system_administration | 3,400 | universal_sft_rlef | A production Ubuntu 22.04 LTS server running Nginx 1.24, PostgreSQL 15, Redis 7, and a Go-based API service behind HAProxy is exhibiting intermittent connection timeouts, high iowait, and sporadic OOM kills during traffic spikes above 800 req/s. The system has LVM on mdadm RAID-10 with 4x NVMe SSDs, 128GB RAM, and cust... | <thought>
Deconstruct objectives: Deliver exact, reproducible diagnostic sequence restricted to listed live tools only (ss, iostat, iotop, perf, bpftrace, journalctl, strace limited to postgres/haproxy, pg_stat_statements). Must include commands, plausible intermediate outputs, and branching root-cause hypotheses. Foll... |
UNIV-AGNT-000001-R01 | UNIV-AGNT-000001 | software_engineering | 3,400 | universal_sft_rlef | You are debugging a production Go microservice that processes financial trades using PostgreSQL with READ COMMITTED isolation. Under sustained 50k TPS load with 200ms network jitter to replicas, duplicate trade executions and lost updates occur during failover. Perform step-by-step root-cause analysis covering transact... | <thought>
Deconstruct: User requires (1) step-by-step RCA covering tx boundaries, advisory locks vs SELECT FOR UPDATE, WAL lag, pool exhaustion under 50k TPS + 200 ms jitter + failover, and (2) production-grade corrected Go implementation with idempotency keys, 2PC emulation, schema migrations, OpenTelemetry instrument... |
UNIV-AGNT-000003-R06 | UNIV-AGNT-000003 | data_engineering | 3,400 | universal_sft_rlef | Design a highly scalable real-time data pipeline capable of processing 500,000 events per second from IoT sensors worldwide. Utilize Apache Kafka for ingestion, Apache Flink for stream processing with complex event processing for anomaly detection, and AWS Redshift for the serving layer. Incorporate data quality checks... | <thought>
Deconstruct objectives: Deliver production-grade IoT pipeline at 500k eps with exact stack (Kafka/MSK, Flink CEP, Redshift), plus quality/GDPR/anonymization, schema evolution, auto-scaling, stateful processing, checkpointing, tuning, DR, and AWS cost model. Map constraints: throughput demands high partition c... |
UNIV-AGNT-000002-R07 | UNIV-AGNT-000002 | system_administration | 3,400 | universal_sft_rlef | A production Ubuntu 22.04 LTS server running Nginx 1.24, PostgreSQL 15, Redis 7, and a Go-based API service behind HAProxy is exhibiting intermittent connection timeouts, high iowait, and sporadic OOM kills during traffic spikes above 800 req/s. The system has LVM on mdadm RAID-10 with 4x NVMe SSDs, 128GB RAM, and cust... | <thought>
**Deconstructing the Problem:**
- Server: Ubuntu 22.04 LTS, Nginx 1.24, PostgreSQL 15, Redis 7, Go API, HAProxy
- Symptoms: Intermittent connection timeouts, high iowait, sporadic OOM kills during traffic spikes (>800 req/s)
- Hardware: 4x NVMe SSDs in RAID-10 (mdadm), LVM, 128GB RAM
- Constraints: Live debug... |
UNIV-AGNT-000002-R04 | UNIV-AGNT-000002 | system_administration | 3,400 | universal_sft_rlef | A production Ubuntu 22.04 LTS server running Nginx 1.24, PostgreSQL 15, Redis 7, and a Go-based API service behind HAProxy is exhibiting intermittent connection timeouts, high iowait, and sporadic OOM kills during traffic spikes above 800 req/s. The system has LVM on mdadm RAID-10 with 4x NVMe SSDs, 128GB RAM, and cust... | <thought>
The scenario presents a complex production system with multiple potential failure points. Let me systematically deconstruct:
**System Architecture:**
- Ubuntu 22.04 LTS
- Nginx 1.24 (reverse proxy/static content)
- HAProxy (load balancer)
- Go API service (backend)
- PostgreSQL 15 (database)
- Redis 7 (cachi... |
UNIV-AGNT-000003-R09 | UNIV-AGNT-000003 | data_engineering | 3,400 | universal_sft_rlef | Design a highly scalable real-time data pipeline capable of processing 500,000 events per second from IoT sensors worldwide. Utilize Apache Kafka for ingestion, Apache Flink for stream processing with complex event processing for anomaly detection, and AWS Redshift for the serving layer. Incorporate data quality checks... | <thought>
This is a complex, multi-faceted request for a high-scale IoT data pipeline. I need to break it down into logical components and address each systematically.
First, I'll design the architecture logically: IoT devices → MQTT/HTTP endpoint → Kafka (ingestion) → Flink (stream processing) → Redshift (serving). I... |
UNIV-AGNT-000004-R01 | UNIV-AGNT-000004 | swarm_orchestration | 3,400 | universal_sft_rlef | Orchestrate a heterogeneous swarm of 60 UAVs and 25 UGVs for persistent surveillance and dynamic target interception across a 2 km^{2} GPS-denied urban disaster zone containing collapsing structures, electromagnetic interference, and variable 15-30 km/h crosswinds. Design a fully decentralized three-layer architecture:... | <thought>
Deconstruct objectives: Deliver fully decentralized 3-layer swarm control for 85 heterogeneous agents (60 UAV + 25 UGV) performing persistent surveillance + dynamic target interception in 2 km^{2} GPS-denied urban disaster zone under EMI, structural collapse, and 15-30 km/h crosswinds. Constraints: no central... |
UNIV-AGNT-000002-R02 | UNIV-AGNT-000002 | system_administration | 3,400 | universal_sft_rlef | A production Ubuntu 22.04 LTS server running Nginx 1.24, PostgreSQL 15, Redis 7, and a Go-based API service behind HAProxy is exhibiting intermittent connection timeouts, high iowait, and sporadic OOM kills during traffic spikes above 800 req/s. The system has LVM on mdadm RAID-10 with 4x NVMe SSDs, 128GB RAM, and cust... | <thought>
This is a complex production system with multiple interacting components. Let me decompose systematically:
1. **Symptoms**: Intermittent connection timeouts, high iowait, sporadic OOM kills during traffic spikes >800 req/s
2. **Stack**: HAProxy → Go API → PostgreSQL 15 + Redis 7 on Ubuntu 22.04
3. **Storage*... |
UNIV-AGNT-000002-R10 | UNIV-AGNT-000002 | system_administration | 3,400 | universal_sft_rlef | A production Ubuntu 22.04 LTS server running Nginx 1.24, PostgreSQL 15, Redis 7, and a Go-based API service behind HAProxy is exhibiting intermittent connection timeouts, high iowait, and sporadic OOM kills during traffic spikes above 800 req/s. The system has LVM on mdadm RAID-10 with 4x NVMe SSDs, 128GB RAM, and cust... | <thought>
The scenario describes a high-throughput LAMP-like stack (HAProxy → Nginx → Go API → PostgreSQL/Redis) on Ubuntu 22.04 with RAID-10 NVMe. Symptoms are timeouts, iowait, and OOM during >800 req/s spikes. I need to construct a strictly tool-constrained diagnostic sequence using only ss, iostat, iotop, perf, bpf... |
End of preview. Expand in Data Studio
⚙️ Universal Frontier: Systems & Concurrency Invariants (Preview)
Hardware-Verified Low-Level Systems, Formal Invariants & Distributed Memory Alignment
-orange.svg){kind=icon}
🏛️ Executive Summary
Contains 92,742 audited high-token-density training trajectories (~350M tokens) focusing strictly on low-level systems programming, formal verification, lock-free concurrency, and kernel invariants. Proven on 8× NVIDIA H100 SXM5 hardware to double base model pass rates on hard SWE-bench and kernel invariants (+72 Elo Gain, 25.2% latency speedup).
This Repository: Hosts a curated 500-sample preview demonstrating formal Raft/Paxos quorums, C++20 lockless SPMC ring buffers, and Linux kernel eBPF XDP verifier bounds.
🛒 License Full 93k Dataset + LoRA 1 Weights ($4,500 USD) on yaeger.info
from datasets import load_dataset
dataset = load_dataset("swarmagents/ufr-systems-concurrency-93k-preview")
print(dataset["train"][0])
- Downloads last month
- 20