Round 4 breakdown

scoredchallengerSubmission

Round score

0.2808

weighted tier speedup after failure cap and deduction

Eligible requests

trusted baseline exclusions removed

Failed eligible requests

each request counted once

Raw weighted speedup

+30.01%

before failure cap and deduction

Score calculation

Raw weighted speedup

+30.01%

Eligible speedup

+30.01%

Request failures cap speed credit at zero: min(0.300136, 0) = 0.300136. Negative speedups remain negative.

Failed requests

6 / 31 (19.35%)

Deduction

0.1 × (6 / 31) = 0.019355

Final score

0.300136 - 0.019355 = 0.280782

Tier completion

Each duration is the median across repetitions, from the group start until the tier's last request finishes. This includes client queueing. Weighted speedup = sum of weight × (1 − candidate / baseline).

Input tier	Eligible requests	Weight	Baseline	Candidate	Speedup	Contribution

The round score uses weighted completion times for all eligible work in each tier. Per-request speedups below are diagnostics, not score contributions. Only trusted baseline exclusions are removed; candidate failures remain in the eligible count and incur the configured failure penalty.

Per-request diagnostics

31 prompts

Prompt	Diagnostic speedup	Baseline	Candidate	Aligned output	Diagnostic / failure
hf-011	—	—	—	3000	Per-token timing unavailable (not a request failure)
hf-002	—	—	—	2239	Per-token timing unavailable (not a request failure)
hf-014	—	—	—

Request trace

31 requests

Request	TTFT	Total	Tokens	Slowest gap	Stream
hf-000	3.769s	79.372s	5120	43.08ms	2652 gaps for 5120 tokens
hf-001	3.640s	84.578s	5120	142.53ms	3009 gaps for 5120 tokens
hf-002	3.926s	68.556s

Observed request concurrency

Occupied client slots include protocol completion and final drain. Mean occupancy is weighted by elapsed time; it can be below the requested cap.

Repetition	Group start	Requested	Effective cap	Peak	Mean
1	0.000s	C32	31	31	25.15
2	0.000s	C32	31	31	25.44
3	0.000s	C32	31	31	25.61

Per-prompt repetition checks

Each latency-median response is compared with the lowest valid opening-baseline ratio for the same prompt. Drops above the recorded limit fail; absolute checks also apply. These checks detect repetition, not overall writing quality.

Prompt	Candidate char-16

Round 4 breakdown

Input tier

Eligible requests

Weight

Baseline

Candidate

Speedup

Contribution

Prompt

Diagnostic speedup

Baseline

Candidate

Aligned output

Diagnostic / failure

hf-011

—

3000

Per-token timing unavailable (not a request failure)

hf-002

—

2239

Per-token timing unavailable (not a request failure)

hf-014

—

Request

TTFT

Total

Tokens

Slowest gap

Stream

hf-000

3.769s

79.372s

5120

43.08ms

2652 gaps for 5120 tokens

hf-001

3.640s

84.578s

5120

142.53ms

3009 gaps for 5120 tokens

hf-002

3.926s

68.556s

Repetition

Group start

Requested

Effective cap

Peak

Mean

0.000s

C32

25.15

0.000s

C32

25.44

0.000s

C32

25.61

hf-000	0.6976	0.6388	-0.0587	0.1000	Passed
hf-001	0.5294	0.7296	0.2002	0.1000	Failed distinct 16-gram ratio 0.529 is more than 0.10 below baseline 0.730 over 24070 chars
hf-002	0.9506	0.9361	-0.0144	0.1000	Passed
hf-003	0.7518	0.7174	-0.0344	0.1000	Passed
hf-004	0.8153	0.8013	-0.0139	0.1000	Passed
hf-005	0.8388	0.6697	-0.1692	0.1000	Passed
hf-006	0.7600	0.6384	-0.1216	0.1000	Passed
hf-007	0.8440	0.7552	-0.0888	0.1000	Passed
hf-008	0.8424	0.8194	-0.0230	0.1000	Passed
hf-009	0.9347	0.8994	-0.0353	0.1000	Passed
hf-010	0.8508	0.6767	-0.1741	0.1000	Passed
hf-011	0.7455	0.8140	0.0685	0.1000	Passed
hf-012	0.9084	0.7565	-0.1519	0.1000	Passed
hf-013	0.7802	0.9176	0.1374	0.1000	Failed distinct 16-gram ratio 0.780 is more than 0.10 below baseline 0.918 over 20730 chars
hf-014	0.8317	0.8834	0.0517	0.1000	Passed
hf-015	0.8890	0.7942	-0.0948	0.1000	Passed
hf-016	0.6514	0.8334	0.1820	0.1000	Failed distinct 16-gram ratio 0.651 is more than 0.10 below baseline 0.833 over 24517 chars
hf-017	0.7316	0.7593	0.0277	0.1000	Passed
hf-018	0.7367	0.8888	0.1521	0.1000	Failed distinct 16-gram ratio 0.737 is more than 0.10 below baseline 0.889 over 21389 chars
hf-019	0.8296	0.7306	-0.0991	0.1000	Passed
hf-020	0.5838	0.5574	-0.0264	0.1000	Passed
hf-021	0.7697	0.7405	-0.0292	0.1000	Passed
hf-022	0.8348	0.7670	-0.0677	0.1000	Passed
hf-023	0.9686	0.8767	-0.0919	0.1000	Passed
hf-024	0.7552	0.6588	-0.0964	0.1000	Passed
hf-025	0.9111	0.8808	-0.0303	0.1000	Passed
hf-027	0.6820	0.6940	0.0120	0.1000	Passed
hf-028	0.7314	0.6719	-0.0594	0.1000	Passed
hf-029	0.7597	0.7355	-0.0242	0.1000	Passed
hf-030	0.7148	0.7141	-0.0007	0.1000	Passed
hf-031	0.7584	0.5707	-0.1877	0.1000	Passed

Prompt	Repetition	Temperature	Top p
hf-023	1	0.446593	1
hf-012	1	0.511421	1
hf-016	1	0.724321	1
hf-008	1	0.612374	1
hf-002	1	0.440527	1
hf-018	1	0.72616	1
hf-015	1	0.992364	1
hf-004	1	0.816071	1
hf-003	1	0.313504	1
hf-009	1	1.009705	1
hf-013	1	0.321389	1
hf-019	1	0.460991	1
hf-010	1	0.782959	1
hf-022	1	0.90875	1
hf-029	1	0.498641	1
hf-006	1	0.832691	1
hf-014	1	0.91642	1
hf-024	1	0.897191	1
hf-005	1	0.528863	1
hf-011	1	0.76435	1
hf-021	1	0.56603	1
hf-000	1	0.567334	1
hf-025	1	0.397412	1
hf-031	1	0.506728	1
hf-027	1	0.44245	1
hf-017	1	0.397818	1
hf-007	1	0.251762	1
hf-020	1	0.245154	1
hf-030	1	0.321942	1
hf-028	1	0.243526	1
hf-001	1	0.143471	1
hf-009	2	1.009705	1
hf-012	2	0.511421	1
hf-007	2	0.251762	1
hf-023	2	0.446593	1
hf-019	2	0.460991	1
hf-014	2	0.91642	1
hf-015	2	0.992364	1
hf-022	2	0.90875	1
hf-010	2	0.782959	1
hf-025	2	0.397412	1
hf-004	2	0.816071	1
hf-024	2	0.897191	1
hf-003	2	0.313504	1
hf-011	2	0.76435	1
hf-006	2	0.832691	1
hf-016	2	0.724321	1
hf-018	2	0.72616	1
hf-008	2	0.612374	1
hf-021	2	0.56603	1
hf-000	2	0.567334	1
hf-029	2	0.498641	1
hf-031	2	0.506728	1
hf-005	2	0.528863	1
hf-002	2	0.440527	1
hf-017	2	0.397818	1
hf-027	2	0.44245	1
hf-030	2	0.321942	1
hf-013	2	0.321389	1
hf-020	2	0.245154	1
hf-028	2	0.243526	1
hf-001	2	0.143471	1
hf-006	3	0.832691	1
hf-008	3	0.612374	1
hf-009	3	1.009705	1
hf-024	3	0.897191	1
hf-007	3	0.251762	1
hf-010	3	0.782959	1
hf-015	3	0.992364	1
hf-020	3	0.245154	1
hf-014	3	0.91642	1
hf-023	3	0.446593	1
hf-001	3	0.143471	1
hf-019	3	0.460991	1
hf-002	3	0.440527	1
hf-005	3	0.528863	1
hf-004	3	0.816071	1
hf-011	3	0.76435	1
hf-025	3	0.397412	1
hf-016	3	0.724321	1
hf-022	3	0.90875	1
hf-018	3	0.72616	1
hf-027	3	0.44245	1
hf-031	3	0.506728	1
hf-021	3	0.56603	1
hf-000	3	0.567334	1
hf-012	3	0.511421	1
hf-029	3	0.498641	1
hf-017	3	0.397818	1
hf-013	3	0.321389	1
hf-030	3	0.321942	1
hf-003	3	0.313504	1
hf-028	3	0.243526	1

Round 4 breakdown

Score calculation

Tier completion

Per-request diagnostics

Request trace

Observed request concurrency

Per-prompt repetition checks

Round 4 breakdown

Round 4 breakdown

Score calculation

Tier completion

Per-request diagnostics

Request trace

Observed request concurrency

Per-prompt repetition checks

Replay sampling

Correctness

SLA replay

Engine