ninfer-16g-5070ti-5080-5090-qwen3.8-27b-gsq-rco/eval/configs/qwen3_6_27b_reasoning.yaml at main · Ryan-gsq/ninfer-16g-5070ti-5080-5090-qwen3.8-27b-gsq-rco · 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
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
schema_version: 1

targets:
  qwen27_api:
    protocol: openai_chat
    base_url: http://127.0.0.1:18080/v1
    model: qwen3.6-27b
    max_concurrency: 1
    request:
      timeout_seconds: 86400
      retries: 2
      retry_interval_seconds: 10

suites:
  reasoning_full:
    jobs:
      - id: aime25
        backend: evalscope
        dataset: aime25
        target: qwen27_api
        max_concurrency: 1
        generation: &reasoning_generation
          temperature: 0.6
          top_p: 0.95
          top_k: 20
          presence_penalty: 1.0
          frequency_penalty: 0.0
          max_tokens: 262144
          seed: 42
        backend_args: &aime_backend
          judge_strategy: rule

      - id: aime26
        backend: evalscope
        dataset: aime26
        target: qwen27_api
        max_concurrency: 1
        generation: *reasoning_generation
        backend_args: *aime_backend

      - id: gpqa_diamond
        backend: evalscope
        dataset: gpqa_diamond
        target: qwen27_api
        max_concurrency: 1
        generation: *reasoning_generation
        backend_args:
          few_shot_num: 0
          judge_strategy: rule

runtime:
  max_parallel_jobs: 1
  runs_dir: eval/runs
  progress:
    enabled: true
    refresh_seconds: 1
    heartbeat_seconds: 30
  samples:
    retention: all