type stringclasses 2
values | harness stringclasses 1
value | id stringclasses 1
value | name stringclasses 1
value | message dict |
|---|---|---|---|---|
session | trackio | agent_run | Agent repro run 2026-08-03 00:13 UTC | null |
message | null | null | null | {
"role": "system",
"content": "Event",
"timestamp": 1785716004958
} |
message | null | null | null | {
"role": "assistant",
"content": "[Claim 1 — Token-batch pipeline scheduling overlap (C1)]\n# Claim 1 — Token-batch pipeline scheduling overlap (C1)\n\n**Claim:** PipeSD overlaps draft-token generation + communication via token-batch pipeline scheduling (DP-optimized), yielding latency reduction vs vanilla sequent... |
message | null | null | null | {
"role": "assistant",
"content": "[Claim 2 Fix — PipeSD Cloud-Edge Throughput Speedups]\n# Claim 2 Fix — PipeSD Cloud-Edge Throughput Speedups\n\n**Paper:** 1ebAvNphi7\n**Claim (index 2):** \"On a real-world cloud-edge testbed, PipeSD achieves 1.16x–2.16x average TPT speedups over Vanilla, HSL, and EdgeLLM baselin... |
message | null | null | null | {
"role": "assistant",
"content": "[Claim 2 — Dual-threshold NAV acceptance (C2)]\n# Claim 2 — Dual-threshold NAV acceptance (C2)\n\n**Claim:** NAV dual-threshold acceptance (cumulative sequence confidence + single-token confidence) improves accepted-token throughput relative to a single-threshold policy.\n\n## Loc... |
message | null | null | null | {
"role": "assistant",
"content": "[Claim 3 — Avg TPT speedups 1.16x–2.16x over baselines (C3, Table 1)]\n# Claim 3 — Avg TPT speedups 1.16x–2.16x over baselines (C3, Table 1)\n\n**Claim:** On a real cloud-edge testbed, PipeSD achieves 1.16×–2.16× average throughput speedups over Vanilla / HSL / EdgeLLM.\n\n## Stat... |
message | null | null | null | {
"role": "assistant",
"content": "[Claim 4 — Cloud energy per 100 accepted tokens reduced 14.3%–25.3% (C4, Table 2)]\n# Claim 4 — Cloud energy per 100 accepted tokens reduced 14.3%–25.3% (C4, Table 2)\n\n**Claim:** In Scenario 1, PipeSD reduces cloud energy per 100 accepted tokens by 14.3%–25.3% relative to Vanill... |
message | null | null | null | {
"role": "assistant",
"content": "[Claim 5 — Speedup maintained across bandwidth levels on HumanEval, Scenario 1 (C5, Figure 5)]\n# Claim 5 — Speedup maintained across bandwidth levels on HumanEval, Scenario 1 (C5, Figure 5)\n\n**Claim:** PipeSD maintains its latency advantage across a sweep of communication bandw... |
message | null | null | null | {
"role": "assistant",
"content": "[Claim 6 — Ablations: improvements from token-batch scheduling + dual-threshold NAV (C6, Table 6)]\n# Claim 6 — Ablations: improvements from token-batch scheduling + dual-threshold NAV (C6, Table 6)\n\n**Claim:** Paper Table 6 shows ablation gains from (1) token-batch pipeline sch... |
message | null | null | null | {
"role": "assistant",
"content": "[Overview — PipeSD (1ebAvNphi7) mechanism validation]\n# Overview — PipeSD (1ebAvNphi7) mechanism validation\n\n**Paper:** PipeSD: An Efficient Cloud-Edge Collaborative Pipeline Inference Framework with Speculative Decoding\n\n**Local logbook slug:** `1ebAvNphi7`\n\nThis local Tra... |
message | null | null | null | {
"role": "system",
"content": "Event",
"timestamp": 1785716004958
} |
- Downloads last month
- 25