Dynavera/benchmarks/results_2026-03-24_13-28-54.md

Dynavera Benchmark Results

Date: 2026-03-24 13:28:54
Inference endpoint: http://fyp-inference-dev:8001
Repetitions per benchmark: 5

1. GPU Server Health

Field	Value
Status	OK
LLM Ready	True
Embed Ready	True
Health check RTT	51.0 ms

2. Embedding Latency

Query type	Chars	Mean (ms)	Median (ms)	P95 (ms)	Min (ms)	Max (ms)
short	19	95.5	25.1	378.6	23.0	378.6
medium	172	25.7	24.7	29.4	24.3	29.4
long	428	27.5	26.7	32.2	24.8	32.2

3. Semantic Chunking Latency

Input size	Chars	Chunks produced	Latency (ms)
small (~200 c)	200	1	28.4
medium (~2k c)	1810	1	77.0
large (~8k c)	7740	1	206.3

4. LLM Inference Latency

Prompt type	Elapsed (s)	Prompt tokens	Completion tokens	Tok/s
short_qa	1.5	55	69	46.0
progress_summary	1.36	74	71	52.3
curriculum_gen	1.67	79	82	49.0
assessment_gen	5.03	83	235	46.7
knowledge_explanation	9.31	83	496	53.3

Note on end-to-end session time: A full onboarding session invokes multiple sequential inference calls (curriculum generation → knowledge explanation × N modules → assessment generation → progress summary). Total wall-clock time accumulates across all turns plus retrieval and tool-call overhead.

5. Database Statistics

Entity	Count
Organizations	3
Roles	10
Users	12
Training Files (total)	0
Training Files (embedded)	0
Knowledge Chunks (with embeddings)	0
Onboarding Sessions	4

Raw JSON

{
  "health": {
    "status": "OK",
    "llm_ready": true,
    "embed_ready": true,
    "latency_ms": 51.0
  },
  "embeddings": {
    "short": {
      "query_chars": 19,
      "mean_ms": 95.5,
      "median_ms": 25.1,
      "p95_ms": 378.6,
      "min_ms": 23.0,
      "max_ms": 378.6
    },
    "medium": {
      "query_chars": 172,
      "mean_ms": 25.7,
      "median_ms": 24.7,
      "p95_ms": 29.4,
      "min_ms": 24.3,
      "max_ms": 29.4
    },
    "long": {
      "query_chars": 428,
      "mean_ms": 27.5,
      "median_ms": 26.7,
      "p95_ms": 32.2,
      "min_ms": 24.8,
      "max_ms": 32.2
    }
  },
  "chunking": {
    "small  (~200 c)": {
      "chars": 200,
      "chunks_produced": 1,
      "latency_ms": 28.4
    },
    "medium (~2k c)": {
      "chars": 1810,
      "chunks_produced": 1,
      "latency_ms": 77.0
    },
    "large  (~8k c)": {
      "chars": 7740,
      "chunks_produced": 1,
      "latency_ms": 206.3
    }
  },
  "llm": {
    "short_qa": {
      "elapsed_s": 1.5,
      "prompt_tokens": 55,
      "completion_tokens": 69,
      "tokens_per_sec": 46.0,
      "response_preview": "A Kubernetes pod is a logical host for one or more containers, providing a shared network namespace,"
    },
    "progress_summary": {
      "elapsed_s": 1.36,
      "prompt_tokens": 74,
      "completion_tokens": 71,
      "tokens_per_sec": 52.3,
      "response_preview": "The trainee has made significant progress in their onboarding journey, demonstrating a strong founda"
    },
    "curriculum_gen": {
      "elapsed_s": 1.67,
      "prompt_tokens": 79,
      "completion_tokens": 82,
      "tokens_per_sec": 49.0,
      "response_preview": "[   \"Module 1: Introduction to Backend Services and Infrastructure\",   \"Module 2: Designing and Impl"
    },
    "assessment_gen": {
      "elapsed_s": 5.03,
      "prompt_tokens": 83,
      "completion_tokens": 235,
      "tokens_per_sec": 46.7,
      "response_preview": "```json [   {     \"question\": \"What is the primary purpose of a Continuous Integration (CI) pipeline"
    },
    "knowledge_explanation": {
      "elapsed_s": 9.31,
      "prompt_tokens": 83,
      "completion_tokens": 496,
      "tokens_per_sec": 53.3,
      "response_preview": "**Git Branching Strategy Best Practices**  As a new engineer, understanding a Git branching strategy"
    }
  },
  "database": {
    "organizations": 3,
    "roles": 10,
    "users": 12,
    "training_files_total": 0,
    "training_files_embedded": 0,
    "knowledge_chunks_with_embeddings": 0,
    "onboarding_sessions": 4
  },
  "retrieval": {
    "skipped": "No embedded chunks found in database."
  }
}