MinerU-ROCm/model_card.windows-hip.json at main · AIwork4me/MinerU-ROCm · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
{
  "model_id": "mineru2.5",
  "platform": "windows-hip",
  "backend": "vlm-llamacpp",
  "backend_detail": {
    "server": "llama.cpp HIP (build b9892, commit ee445f93)",
    "quantization": "Q8_0 (text + mmproj)",
    "context": 8192,
    "parallel_slots": 1,
    "sampling": "greedy (temperature=0, top_k=1)",
    "visual_tokens": "1024-1600",
    "flash_attention": "enabled"
  },
  "hardware": {
    "gpu": "AMD Radeon 8060S (Strix Halo, gfx1151)",
    "driver": "AMD HIP SDK (ROCm 7.2.1 via Windows HIP)",
    "memory": "128 GB unified (CPU+GPU shared)"
  },
  "evaluation": {
    "dataset": "OmniDocBench v1.6 (revision 2b161d0)",
    "pages": 1651,
    "overall": 95.46,
    "text_edit_dist": 0.03734,
    "formula_cdm": 0.97008,
    "table_teds": 0.93105,
    "reading_order_edit_dist": 0.1225,
    "ok_count": 1649,
    "fail_count": 2,
    "fallback_count": 0,
    "cdm_samples": 2352,
    "cdm_timeout": 0,
    "teds_samples": 665,
    "teds_timeout": 0
  },
  "comparison": {
    "baseline": "Linux vLLM on ROCm (1,651 pages)",
    "baseline_overall": 95.46,
    "delta_overall": 0.0
  },
  "bundle": "results/omnidocbench/v16/windows-hip/",
  "source_commit": "274c4fe",
  "community_badge": "windows-hip-verified"
}