TensorRT-LLM reporting negative value for throughput and no zero for inter token latency

AI & Data Science AI Foundation Models and Endpoints
1

I am running the gptBenchmark provided by TensorRT-LLM on Mixtral 8x7B on H100 GPU. The results I got reports throughput as negative , inter token latency is reported as zero. Why so ?

Below is a sample report :

[BENCHMARK] num_samples 64
[BENCHMARK] num_error_samples 0

[BENCHMARK] num_samples 64
[BENCHMARK] total_latency(ms) 24533.64
[BENCHMARK] seq_throughput(seq/sec) 2.61
[BENCHMARK] token_throughput(token/sec) -28692.68

[BENCHMARK] avg_sequence_latency(ms) 22124.27
[BENCHMARK] max_sequence_latency(ms) 24506.07
[BENCHMARK] min_sequence_latency(ms) 18681.99
[BENCHMARK] p99_sequence_latency(ms) 24506.07
[BENCHMARK] p90_sequence_latency(ms) 24207.51
[BENCHMARK] p50_sequence_latency(ms) 22450.42

[BENCHMARK] avg_time_to_first_token(ms) 7186.20
[BENCHMARK] max_time_to_first_token(ms) 15130.73
[BENCHMARK] min_time_to_first_token(ms) 194.69
[BENCHMARK] p99_time_to_first_token(ms) 15130.73
[BENCHMARK] p90_time_to_first_token(ms) 13521.66
[BENCHMARK] p50_time_to_first_token(ms) 6768.20

[BENCHMARK] avg_inter_token_latency(ms) 0.00
[BENCHMARK] max_inter_token_latency(ms) 0.00
[BENCHMARK] min_inter_token_latency(ms) 0.00
[BENCHMARK] p99_inter_token_latency(ms) 0.00
[BENCHMARK] p90_inter_token_latency(ms) 0.00
[BENCHMARK] p50_inter_token_latency(ms) 0.00