Zum Inhalt

tests/test_results.py

  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
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
def test_evalresponse_reasoning_text_defaults_empty():
    from touchstone.results import EvalResponse

    r = EvalResponse(
        pack_id="p",
        pack_version=1,
        machine="M",
        model="m",
        quant="",
        engine="e",
        engine_version="0",
        variant="none",
        category="A",
        prompt_id="A1",
        repeat=0,
        response_text="",
        content_empty=True,
        ttft_s=0.0,
        decode_tps=0.0,
        prefill_tps=0.0,
        e2e_s=0.0,
        prompt_tokens=0,
        completion_tokens=0,
        is_cold_start=False,
        power_source="ac",
        peak_rss_mb=None,
        sys_used_mb=None,
        mem_pressure_max="",
        throttled=False,
        ok=True,
        error="",
        seed=42,
        t_start=0.0,
        t_end=0.0,
    )
    assert r.reasoning_text == ""
    assert r.as_dict()["reasoning_text"] == ""
    # model-delta fields default safely and serialize
    assert r.sys_used_baseline_mb is None
    assert r.sys_used_delta_mb is None
    d = r.as_dict()
    assert d["sys_used_baseline_mb"] is None
    assert d["sys_used_delta_mb"] is None


def test_evalresponse_reasoning_timing_defaults_safe():
    from touchstone.results import EvalResponse

    r = EvalResponse(
        pack_id="p",
        pack_version=1,
        machine="M",
        model="m",
        quant="",
        engine="e",
        engine_version="0",
        variant="none",
        category="A",
        prompt_id="A1",
        repeat=0,
        response_text="",
        content_empty=True,
        ttft_s=0.0,
        decode_tps=0.0,
        prefill_tps=0.0,
        e2e_s=0.0,
        prompt_tokens=0,
        completion_tokens=0,
        is_cold_start=False,
        power_source="ac",
        peak_rss_mb=None,
        sys_used_mb=None,
        mem_pressure_max="",
        throttled=False,
        ok=True,
        error="",
        seed=42,
        t_start=0.0,
        t_end=0.0,
    )
    # new reasoning-timing fields default safely and serialize
    assert r.reasoning_duration_s == 0.0
    assert r.reasoning_tps == 0.0
    assert r.reasoning_completion_tokens == 0
    d = r.as_dict()
    assert d["reasoning_duration_s"] == 0.0
    assert d["reasoning_tps"] == 0.0
    assert d["reasoning_completion_tokens"] == 0


def test_evalresponse_carries_reasoning_timing_when_set():
    from touchstone.results import EvalResponse

    r = EvalResponse(
        pack_id="p",
        pack_version=1,
        machine="M",
        model="m",
        quant="",
        engine="e",
        engine_version="0",
        variant="none",
        category="A",
        prompt_id="A1",
        repeat=0,
        response_text="x",
        content_empty=False,
        ttft_s=0.0,
        decode_tps=0.0,
        prefill_tps=0.0,
        e2e_s=0.0,
        prompt_tokens=0,
        completion_tokens=0,
        is_cold_start=False,
        power_source="ac",
        peak_rss_mb=None,
        sys_used_mb=None,
        mem_pressure_max="",
        throttled=False,
        ok=True,
        error="",
        seed=42,
        t_start=0.0,
        t_end=0.0,
        reasoning_duration_s=2.5,
        reasoning_tps=40.0,
        reasoning_completion_tokens=100,
    )
    d = r.as_dict()
    assert d["reasoning_duration_s"] == 2.5
    assert d["reasoning_tps"] == 40.0
    assert d["reasoning_completion_tokens"] == 100


def test_evalresponse_carries_model_delta_when_set():
    from touchstone.results import EvalResponse

    r = EvalResponse(
        pack_id="p",
        pack_version=1,
        machine="M",
        model="m",
        quant="",
        engine="e",
        engine_version="0",
        variant="none",
        category="A",
        prompt_id="A1",
        repeat=0,
        response_text="x",
        content_empty=False,
        ttft_s=0.0,
        decode_tps=0.0,
        prefill_tps=0.0,
        e2e_s=0.0,
        prompt_tokens=0,
        completion_tokens=0,
        is_cold_start=False,
        power_source="ac",
        peak_rss_mb=None,
        sys_used_mb=52000.0,
        mem_pressure_max="",
        throttled=False,
        ok=True,
        error="",
        seed=42,
        t_start=0.0,
        t_end=0.0,
        sys_used_baseline_mb=40000.0,
        sys_used_delta_mb=12000.0,
    )
    assert r.as_dict()["sys_used_delta_mb"] == 12000.0
    assert r.as_dict()["sys_used_baseline_mb"] == 40000.0