tarekziade HF Staff commited on
Commit
ca9dfcb
·
verified ·
1 Parent(s): 64c7620

Review cloud MIC baselines arm64-GLM5Next

Browse files

Measured and independently validated on CPU and CUDA. Review reasons: {"cpu/prototypes/GLM5Next/GLM-5.3-Flash-mic/bfloat16": {"reason": "New or unchanged measured reference", "sha256": "d0c591a9e82d33d2b7b2aa21ffbc3265fef1bf087a44e4d240c075cfbd812525", "verdict": "new"}, "cpu/prototypes/GLM5Next/GLM-5.3-Flash-mic/default": {"reason": "New or unchanged measured reference", "sha256": "d316f7b72ce605899d12fac89ff9f17597c600d1816bf40e766e38d7bcd1527b", "verdict": "new"}, "cpu/prototypes/GLM5Next/GLM-5.3-Flash-mic/eager": {"reason": "New or unchanged measured reference", "sha256": "e1bb687f7993f1922d121aa2928018dca9141b18388fdc4ea36171b19b9fb235", "verdict": "new"}}

integrations/transformers/expectations/cpu-arm64/GLM-5.3-Flash-mic.bfloat16.json ADDED
@@ -0,0 +1,78 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "recorded_with": {
3
+ "contract": "huggingface-mic/GLM-5.3-Flash-mic",
4
+ "contract_name": "GLM5Next",
5
+ "contract_revision": "64c7620a37be59fb17eabca05f932b7f2d78b502",
6
+ "framework": "transformers",
7
+ "variant": "bfloat16",
8
+ "target": "GLM-5.3-Flash-mic",
9
+ "kind": "fixture",
10
+ "revision": "64c7620a37be59fb17eabca05f932b7f2d78b502",
11
+ "source_model": "zai-org/GLM-5.3-Flash",
12
+ "weights": "random",
13
+ "fixture_repo": "huggingface-mic/GLM-5.3-Flash-mic",
14
+ "environment": {
15
+ "python": "3.11.14",
16
+ "transformers": "5.23.0.dev0",
17
+ "sentence-transformers": null,
18
+ "torch": "2.15.0",
19
+ "torchvision": "0.29.2",
20
+ "torchaudio": "2.11.1",
21
+ "tokenizers": "0.23.3",
22
+ "numpy": "2.3.5",
23
+ "pillow": "13.0.0",
24
+ "accelerate": "1.15.0",
25
+ "safetensors": "0.8.0",
26
+ "huggingface-hub": "2.4.0",
27
+ "sentencepiece": "0.2.2",
28
+ "protobuf": "7.36.2",
29
+ "librosa": "0.11.0",
30
+ "soundfile": "0.14.0",
31
+ "scipy": "1.17.1"
32
+ },
33
+ "seconds": 6.61,
34
+ "environment_differs": {
35
+ "torch": "2.15.0 (pinned 2.14.0)",
36
+ "torchvision": "0.29.2 (pinned 0.29.1)",
37
+ "torchaudio": "2.11.1 (pinned 2.11.0)",
38
+ "pillow": "13.0.0 (pinned 12.3.0)",
39
+ "huggingface-hub": "2.4.0 (pinned 2.3.0)"
40
+ },
41
+ "platform": "cpu-arm64",
42
+ "template_date": "2026-01-01"
43
+ },
44
+ "outputs": {
45
+ "runtime": {
46
+ "device": "cpu",
47
+ "dtype": "bfloat16",
48
+ "attn_implementation": "sdpa",
49
+ "attn_implementations": {
50
+ "text_config": "sdpa",
51
+ "vision_config": "sdpa"
52
+ }
53
+ },
54
+ "predicted_tokens": [
55
+ "\u0120\"\"\u010a\u010a",
56
+ "VERIFY"
57
+ ],
58
+ "answer_logits": [
59
+ [
60
+ 0.921875,
61
+ -0.326171875
62
+ ],
63
+ [
64
+ 0.11083984375,
65
+ 1.03125
66
+ ]
67
+ ],
68
+ "training": {
69
+ "loss": 11.998385429382324,
70
+ "gradients": {
71
+ "model.language_model.layers.0.self_attn.q_proj.weight": true,
72
+ "model.language_model.layers.0.mlp.gate_proj.weight": true,
73
+ "model.visual.blocks.0.attn.qkv.weight": true
74
+ },
75
+ "updated": true
76
+ }
77
+ }
78
+ }
integrations/transformers/expectations/cpu-arm64/GLM-5.3-Flash-mic.eager.json ADDED
@@ -0,0 +1,78 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "recorded_with": {
3
+ "contract": "huggingface-mic/GLM-5.3-Flash-mic",
4
+ "contract_name": "GLM5Next",
5
+ "contract_revision": "64c7620a37be59fb17eabca05f932b7f2d78b502",
6
+ "framework": "transformers",
7
+ "variant": "eager",
8
+ "target": "GLM-5.3-Flash-mic",
9
+ "kind": "fixture",
10
+ "revision": "64c7620a37be59fb17eabca05f932b7f2d78b502",
11
+ "source_model": "zai-org/GLM-5.3-Flash",
12
+ "weights": "random",
13
+ "fixture_repo": "huggingface-mic/GLM-5.3-Flash-mic",
14
+ "environment": {
15
+ "python": "3.11.14",
16
+ "transformers": "5.23.0.dev0",
17
+ "sentence-transformers": null,
18
+ "torch": "2.15.0",
19
+ "torchvision": "0.29.2",
20
+ "torchaudio": "2.11.1",
21
+ "tokenizers": "0.23.3",
22
+ "numpy": "2.3.5",
23
+ "pillow": "13.0.0",
24
+ "accelerate": "1.15.0",
25
+ "safetensors": "0.8.0",
26
+ "huggingface-hub": "2.4.0",
27
+ "sentencepiece": "0.2.2",
28
+ "protobuf": "7.36.2",
29
+ "librosa": "0.11.0",
30
+ "soundfile": "0.14.0",
31
+ "scipy": "1.17.1"
32
+ },
33
+ "seconds": 3.06,
34
+ "environment_differs": {
35
+ "torch": "2.15.0 (pinned 2.14.0)",
36
+ "torchvision": "0.29.2 (pinned 0.29.1)",
37
+ "torchaudio": "2.11.1 (pinned 2.11.0)",
38
+ "pillow": "13.0.0 (pinned 12.3.0)",
39
+ "huggingface-hub": "2.4.0 (pinned 2.3.0)"
40
+ },
41
+ "platform": "cpu-arm64",
42
+ "template_date": "2026-01-01"
43
+ },
44
+ "outputs": {
45
+ "runtime": {
46
+ "device": "cpu",
47
+ "dtype": "float32",
48
+ "attn_implementation": "eager",
49
+ "attn_implementations": {
50
+ "text_config": "eager",
51
+ "vision_config": "eager"
52
+ }
53
+ },
54
+ "predicted_tokens": [
55
+ "\u0120\"\"\u010a\u010a",
56
+ "VERIFY"
57
+ ],
58
+ "answer_logits": [
59
+ [
60
+ 0.9239497184753418,
61
+ -0.32489213347435
62
+ ],
63
+ [
64
+ 0.11050514876842499,
65
+ 1.0303713083267212
66
+ ]
67
+ ],
68
+ "training": {
69
+ "loss": 12.002214431762695,
70
+ "gradients": {
71
+ "model.language_model.layers.0.self_attn.q_proj.weight": true,
72
+ "model.language_model.layers.0.mlp.gate_proj.weight": true,
73
+ "model.visual.blocks.0.attn.qkv.weight": true
74
+ },
75
+ "updated": true
76
+ }
77
+ }
78
+ }
integrations/transformers/expectations/cpu-arm64/GLM-5.3-Flash-mic.json ADDED
@@ -0,0 +1,78 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "recorded_with": {
3
+ "contract": "huggingface-mic/GLM-5.3-Flash-mic",
4
+ "contract_name": "GLM5Next",
5
+ "contract_revision": "64c7620a37be59fb17eabca05f932b7f2d78b502",
6
+ "framework": "transformers",
7
+ "variant": "default",
8
+ "target": "GLM-5.3-Flash-mic",
9
+ "kind": "fixture",
10
+ "revision": "64c7620a37be59fb17eabca05f932b7f2d78b502",
11
+ "source_model": "zai-org/GLM-5.3-Flash",
12
+ "weights": "random",
13
+ "fixture_repo": "huggingface-mic/GLM-5.3-Flash-mic",
14
+ "environment": {
15
+ "python": "3.11.14",
16
+ "transformers": "5.23.0.dev0",
17
+ "sentence-transformers": null,
18
+ "torch": "2.15.0",
19
+ "torchvision": "0.29.2",
20
+ "torchaudio": "2.11.1",
21
+ "tokenizers": "0.23.3",
22
+ "numpy": "2.3.5",
23
+ "pillow": "13.0.0",
24
+ "accelerate": "1.15.0",
25
+ "safetensors": "0.8.0",
26
+ "huggingface-hub": "2.4.0",
27
+ "sentencepiece": "0.2.2",
28
+ "protobuf": "7.36.2",
29
+ "librosa": "0.11.0",
30
+ "soundfile": "0.14.0",
31
+ "scipy": "1.17.1"
32
+ },
33
+ "seconds": 3.06,
34
+ "environment_differs": {
35
+ "torch": "2.15.0 (pinned 2.14.0)",
36
+ "torchvision": "0.29.2 (pinned 0.29.1)",
37
+ "torchaudio": "2.11.1 (pinned 2.11.0)",
38
+ "pillow": "13.0.0 (pinned 12.3.0)",
39
+ "huggingface-hub": "2.4.0 (pinned 2.3.0)"
40
+ },
41
+ "platform": "cpu-arm64",
42
+ "template_date": "2026-01-01"
43
+ },
44
+ "outputs": {
45
+ "runtime": {
46
+ "device": "cpu",
47
+ "dtype": "float32",
48
+ "attn_implementation": "sdpa",
49
+ "attn_implementations": {
50
+ "text_config": "sdpa",
51
+ "vision_config": "sdpa"
52
+ }
53
+ },
54
+ "predicted_tokens": [
55
+ "\u0120\"\"\u010a\u010a",
56
+ "VERIFY"
57
+ ],
58
+ "answer_logits": [
59
+ [
60
+ 0.923949658870697,
61
+ -0.32489219307899475
62
+ ],
63
+ [
64
+ 0.1105053722858429,
65
+ 1.0303711891174316
66
+ ]
67
+ ],
68
+ "training": {
69
+ "loss": 12.001060485839844,
70
+ "gradients": {
71
+ "model.language_model.layers.0.self_attn.q_proj.weight": true,
72
+ "model.language_model.layers.0.mlp.gate_proj.weight": true,
73
+ "model.visual.blocks.0.attn.qkv.weight": true
74
+ },
75
+ "updated": true
76
+ }
77
+ }
78
+ }