Review cloud MIC baselines 7d8449a176b7481f9cb779992b3f8cb3

integrations/transformers/expectations/cpu-x86_64-bf16/vit-mic.bfloat16.json

@@ -1,34 +1,44 @@
 {
  "recorded_with": {
-  "contract": "hf-internal-testing/mic-vit-integration",
-  "contract_name": "vit",
-  "contract_revision": "499b613e58aa698e4a572f6c1265aa55eb60a38e",
+  "contract": "huggingface-mic/vit-mic",
+  "contract_name": "mic-vit",
+  "contract_revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "framework": "transformers",
   "variant": "bfloat16",
   "target": "vit-mic",
   "kind": "fixture",
-  "revision": "2bf504ea8bd2e48d5e5870ea2880f8f79d52e11e",
+  "revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "environment": {
    "python": "3.11.17",
-   "transformers": "5.19.0.dev0",
-   "torch": "2.11.0+cpu",
-   "torchvision": "0.26.0+cpu",
+   "transformers": "5.21.0.dev0",
+   "torch": "2.14.0+cpu",
+   "torchvision": "0.29.1+cpu",
    "torchaudio": "2.11.0+cpu",
-   "tokenizers": "0.23.1",
-   "numpy": "1.26.4",
-   "pillow": "12.2.0",
-   "accelerate": "1.13.0",
+   "tokenizers": "0.23.3",
+   "numpy": "2.3.5",
+   "pillow": "12.3.0",
+   "accelerate": "1.15.0",
    "safetensors": "0.8.0",
-   "huggingface-hub": "1.32.0",
-   "sentencepiece": "0.2.1",
-   "protobuf": "6.33.6",
+   "huggingface-hub": "2.3.0",
+   "sentencepiece": "0.2.2",
+   "protobuf": "7.36.2",
    "librosa": "0.11.0",
-   "soundfile": "0.13.1",
-   "scipy": "1.15.3"
+   "soundfile": "0.14.0",
+   "scipy": "1.17.1"
   },
-  "seconds": 6.81,
+  "seconds": 7.87,
   "environment_differs": {
-   "python": "3.11.17 (pinned 3.10)"
+   "torch": "2.14.0+cpu (pinned 2.11.0)",
+   "torchvision": "0.29.1+cpu (pinned 0.26.0)",
+   "tokenizers": "0.23.3 (pinned 0.23.1)",
+   "numpy": "2.3.5 (pinned 1.26.4)",
+   "pillow": "12.3.0 (pinned 12.2.0)",
+   "accelerate": "1.15.0 (pinned 1.13.0)",
+   "huggingface-hub": "2.3.0 (pinned 1.32.0)",
+   "sentencepiece": "0.2.2 (pinned 0.2.1)",
+   "protobuf": "7.36.2 (pinned 6.33.6)",
+   "soundfile": "0.14.0 (pinned 0.13.1)",
+   "scipy": "1.17.1 (pinned 1.15.3)"
   },
   "platform": "cpu-x86_64-bf16"
  },

integrations/transformers/expectations/cpu-x86_64/vit-mic.eager.json

@@ -1,34 +1,44 @@
 {
  "recorded_with": {
-  "contract": "hf-internal-testing/mic-vit-integration",
-  "contract_name": "vit",
-  "contract_revision": "499b613e58aa698e4a572f6c1265aa55eb60a38e",
+  "contract": "huggingface-mic/vit-mic",
+  "contract_name": "mic-vit",
+  "contract_revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "framework": "transformers",
   "variant": "eager",
   "target": "vit-mic",
   "kind": "fixture",
-  "revision": "2bf504ea8bd2e48d5e5870ea2880f8f79d52e11e",
+  "revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "environment": {
    "python": "3.11.17",
-   "transformers": "5.19.0.dev0",
-   "torch": "2.11.0+cpu",
-   "torchvision": "0.26.0+cpu",
+   "transformers": "5.21.0.dev0",
+   "torch": "2.14.0+cpu",
+   "torchvision": "0.29.1+cpu",
    "torchaudio": "2.11.0+cpu",
-   "tokenizers": "0.23.1",
-   "numpy": "1.26.4",
-   "pillow": "12.2.0",
-   "accelerate": "1.13.0",
+   "tokenizers": "0.23.3",
+   "numpy": "2.3.5",
+   "pillow": "12.3.0",
+   "accelerate": "1.15.0",
    "safetensors": "0.8.0",
-   "huggingface-hub": "1.32.0",
-   "sentencepiece": "0.2.1",
-   "protobuf": "6.33.6",
+   "huggingface-hub": "2.3.0",
+   "sentencepiece": "0.2.2",
+   "protobuf": "7.36.2",
    "librosa": "0.11.0",
-   "soundfile": "0.13.1",
-   "scipy": "1.15.3"
+   "soundfile": "0.14.0",
+   "scipy": "1.17.1"
   },
-  "seconds": 6.7,
+  "seconds": 7.67,
   "environment_differs": {
-   "python": "3.11.17 (pinned 3.10)"
+   "torch": "2.14.0+cpu (pinned 2.11.0)",
+   "torchvision": "0.29.1+cpu (pinned 0.26.0)",
+   "tokenizers": "0.23.3 (pinned 0.23.1)",
+   "numpy": "2.3.5 (pinned 1.26.4)",
+   "pillow": "12.3.0 (pinned 12.2.0)",
+   "accelerate": "1.15.0 (pinned 1.13.0)",
+   "huggingface-hub": "2.3.0 (pinned 1.32.0)",
+   "sentencepiece": "0.2.2 (pinned 0.2.1)",
+   "protobuf": "7.36.2 (pinned 6.33.6)",
+   "soundfile": "0.14.0 (pinned 0.13.1)",
+   "scipy": "1.17.1 (pinned 1.15.3)"
   },
   "platform": "cpu-x86_64"
  },

integrations/transformers/expectations/cpu-x86_64/vit-mic.json

@@ -1,34 +1,44 @@
 {
  "recorded_with": {
-  "contract": "hf-internal-testing/mic-vit-integration",
-  "contract_name": "vit",
-  "contract_revision": "499b613e58aa698e4a572f6c1265aa55eb60a38e",
+  "contract": "huggingface-mic/vit-mic",
+  "contract_name": "mic-vit",
+  "contract_revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "framework": "transformers",
   "variant": "default",
   "target": "vit-mic",
   "kind": "fixture",
-  "revision": "2bf504ea8bd2e48d5e5870ea2880f8f79d52e11e",
+  "revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "environment": {
    "python": "3.11.17",
-   "transformers": "5.19.0.dev0",
-   "torch": "2.11.0+cpu",
-   "torchvision": "0.26.0+cpu",
+   "transformers": "5.21.0.dev0",
+   "torch": "2.14.0+cpu",
+   "torchvision": "0.29.1+cpu",
    "torchaudio": "2.11.0+cpu",
-   "tokenizers": "0.23.1",
-   "numpy": "1.26.4",
-   "pillow": "12.2.0",
-   "accelerate": "1.13.0",
+   "tokenizers": "0.23.3",
+   "numpy": "2.3.5",
+   "pillow": "12.3.0",
+   "accelerate": "1.15.0",
    "safetensors": "0.8.0",
-   "huggingface-hub": "1.32.0",
-   "sentencepiece": "0.2.1",
-   "protobuf": "6.33.6",
+   "huggingface-hub": "2.3.0",
+   "sentencepiece": "0.2.2",
+   "protobuf": "7.36.2",
    "librosa": "0.11.0",
-   "soundfile": "0.13.1",
-   "scipy": "1.15.3"
+   "soundfile": "0.14.0",
+   "scipy": "1.17.1"
   },
-  "seconds": 6.08,
+  "seconds": 7.6,
   "environment_differs": {
-   "python": "3.11.17 (pinned 3.10)"
+   "torch": "2.14.0+cpu (pinned 2.11.0)",
+   "torchvision": "0.29.1+cpu (pinned 0.26.0)",
+   "tokenizers": "0.23.3 (pinned 0.23.1)",
+   "numpy": "2.3.5 (pinned 1.26.4)",
+   "pillow": "12.3.0 (pinned 12.2.0)",
+   "accelerate": "1.15.0 (pinned 1.13.0)",
+   "huggingface-hub": "2.3.0 (pinned 1.32.0)",
+   "sentencepiece": "0.2.2 (pinned 0.2.1)",
+   "protobuf": "7.36.2 (pinned 6.33.6)",
+   "soundfile": "0.14.0 (pinned 0.13.1)",
+   "scipy": "1.17.1 (pinned 1.15.3)"
   },
   "platform": "cpu-x86_64"
  },

integrations/transformers/expectations/cuda/vit-mic.bfloat16.json

@@ -1,31 +1,45 @@
 {
  "recorded_with": {
-  "contract": "hf-internal-testing/mic-vit-integration",
-  "contract_revision": "unpublished",
+  "contract": "huggingface-mic/vit-mic",
+  "contract_name": "mic-vit",
+  "contract_revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "framework": "transformers",
   "variant": "bfloat16",
   "target": "vit-mic",
   "kind": "fixture",
-  "revision": "unpublished",
+  "revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "environment": {
-   "python": "3.10.12",
-   "transformers": "5.19.0.dev0",
-   "torch": "2.11.0+cu128",
-   "torchvision": "0.26.0+cu128",
-   "torchaudio": "2.11.0+cu128",
-   "tokenizers": "0.23.1",
-   "numpy": "1.26.4",
-   "pillow": "12.2.0",
-   "accelerate": "1.13.0",
+   "python": "3.11.18",
+   "transformers": "5.21.0.dev0",
+   "torch": "2.14.0+cu130",
+   "torchvision": "0.29.1+cu130",
+   "torchaudio": "2.11.0+cu130",
+   "tokenizers": "0.23.3",
+   "numpy": "2.4.6",
+   "pillow": "13.0.0",
+   "accelerate": "1.16.0.dev0",
    "safetensors": "0.8.0",
-   "huggingface-hub": "1.32.0",
-   "sentencepiece": "0.2.1",
-   "protobuf": "6.33.6",
+   "huggingface-hub": "2.3.0",
+   "sentencepiece": "0.2.2",
+   "protobuf": "7.36.2",
    "librosa": "0.11.0",
-   "soundfile": "0.13.1",
-   "scipy": "1.15.3"
+   "soundfile": "0.14.0",
+   "scipy": "1.17.1"
+  },
+  "seconds": 6.61,
+  "environment_differs": {
+   "torch": "2.14.0+cu130 (pinned 2.11.0)",
+   "torchvision": "0.29.1+cu130 (pinned 0.26.0)",
+   "tokenizers": "0.23.3 (pinned 0.23.1)",
+   "numpy": "2.4.6 (pinned 1.26.4)",
+   "pillow": "13.0.0 (pinned 12.2.0)",
+   "accelerate": "1.16.0.dev0 (pinned 1.13.0)",
+   "huggingface-hub": "2.3.0 (pinned 1.32.0)",
+   "sentencepiece": "0.2.2 (pinned 0.2.1)",
+   "protobuf": "7.36.2 (pinned 6.33.6)",
+   "soundfile": "0.14.0 (pinned 0.13.1)",
+   "scipy": "1.17.1 (pinned 1.15.3)"
   },
-  "seconds": 4.78,
   "platform": "cuda"
  },
  "outputs": {

integrations/transformers/expectations/cuda/vit-mic.eager.json

@@ -1,31 +1,45 @@
 {
  "recorded_with": {
-  "contract": "hf-internal-testing/mic-vit-integration",
-  "contract_revision": "unpublished",
+  "contract": "huggingface-mic/vit-mic",
+  "contract_name": "mic-vit",
+  "contract_revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "framework": "transformers",
   "variant": "eager",
   "target": "vit-mic",
   "kind": "fixture",
-  "revision": "unpublished",
+  "revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "environment": {
-   "python": "3.10.12",
-   "transformers": "5.19.0.dev0",
-   "torch": "2.11.0+cu128",
-   "torchvision": "0.26.0+cu128",
-   "torchaudio": "2.11.0+cu128",
-   "tokenizers": "0.23.1",
-   "numpy": "1.26.4",
-   "pillow": "12.2.0",
-   "accelerate": "1.13.0",
+   "python": "3.11.18",
+   "transformers": "5.21.0.dev0",
+   "torch": "2.14.0+cu130",
+   "torchvision": "0.29.1+cu130",
+   "torchaudio": "2.11.0+cu130",
+   "tokenizers": "0.23.3",
+   "numpy": "2.4.6",
+   "pillow": "13.0.0",
+   "accelerate": "1.16.0.dev0",
    "safetensors": "0.8.0",
-   "huggingface-hub": "1.32.0",
-   "sentencepiece": "0.2.1",
-   "protobuf": "6.33.6",
+   "huggingface-hub": "2.3.0",
+   "sentencepiece": "0.2.2",
+   "protobuf": "7.36.2",
    "librosa": "0.11.0",
-   "soundfile": "0.13.1",
-   "scipy": "1.15.3"
+   "soundfile": "0.14.0",
+   "scipy": "1.17.1"
+  },
+  "seconds": 6.51,
+  "environment_differs": {
+   "torch": "2.14.0+cu130 (pinned 2.11.0)",
+   "torchvision": "0.29.1+cu130 (pinned 0.26.0)",
+   "tokenizers": "0.23.3 (pinned 0.23.1)",
+   "numpy": "2.4.6 (pinned 1.26.4)",
+   "pillow": "13.0.0 (pinned 12.2.0)",
+   "accelerate": "1.16.0.dev0 (pinned 1.13.0)",
+   "huggingface-hub": "2.3.0 (pinned 1.32.0)",
+   "sentencepiece": "0.2.2 (pinned 0.2.1)",
+   "protobuf": "7.36.2 (pinned 6.33.6)",
+   "soundfile": "0.14.0 (pinned 0.13.1)",
+   "scipy": "1.17.1 (pinned 1.15.3)"
   },
-  "seconds": 4.37,
   "platform": "cuda"
  },
  "outputs": {
@@ -52,22 +66,22 @@
   ],
   "top5_probs": [
    [
-    0.9963154196739197,
-    0.0035723361652344465,
-    1.480734823644525e-07,
-    1.479782127944418e-07,
-    1.4519815749736154e-07
+    0.99631267786026,
+    0.003575016977265477,
+    1.4806560955094028e-07,
+    1.4797343794725748e-07,
+    1.4519096680487564e-07
    ],
    [
     0.9955466985702515,
-    0.000329399888869375,
-    4.8358510866819415e-06,
-    4.792949766851962e-06,
-    4.656029432226205e-06
+    0.00032939674565568566,
+    4.835921117773978e-06,
+    4.7930143409757875e-06,
+    4.656090823118575e-06
    ]
   ],
   "training": {
-   "loss": 0.004077321384102106,
+   "loss": 0.00407868716865778,
    "gradients": {
     "vit.embeddings.cls_token": true,
     "vit.embeddings.position_embeddings": true

integrations/transformers/expectations/cuda/vit-mic.json

@@ -1,31 +1,45 @@
 {
  "recorded_with": {
-  "contract": "hf-internal-testing/mic-vit-integration",
-  "contract_revision": "unpublished",
+  "contract": "huggingface-mic/vit-mic",
+  "contract_name": "mic-vit",
+  "contract_revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "framework": "transformers",
   "variant": "default",
   "target": "vit-mic",
   "kind": "fixture",
-  "revision": "unpublished",
+  "revision": "a6d3f2396e5c8acf9a25d65cb31cf41cca08ba40",
   "environment": {
-   "python": "3.10.12",
-   "transformers": "5.19.0.dev0",
-   "torch": "2.11.0+cu128",
-   "torchvision": "0.26.0+cu128",
-   "torchaudio": "2.11.0+cu128",
-   "tokenizers": "0.23.1",
-   "numpy": "1.26.4",
-   "pillow": "12.2.0",
-   "accelerate": "1.13.0",
+   "python": "3.11.18",
+   "transformers": "5.21.0.dev0",
+   "torch": "2.14.0+cu130",
+   "torchvision": "0.29.1+cu130",
+   "torchaudio": "2.11.0+cu130",
+   "tokenizers": "0.23.3",
+   "numpy": "2.4.6",
+   "pillow": "13.0.0",
+   "accelerate": "1.16.0.dev0",
    "safetensors": "0.8.0",
-   "huggingface-hub": "1.32.0",
-   "sentencepiece": "0.2.1",
-   "protobuf": "6.33.6",
+   "huggingface-hub": "2.3.0",
+   "sentencepiece": "0.2.2",
+   "protobuf": "7.36.2",
    "librosa": "0.11.0",
-   "soundfile": "0.13.1",
-   "scipy": "1.15.3"
+   "soundfile": "0.14.0",
+   "scipy": "1.17.1"
+  },
+  "seconds": 6.47,
+  "environment_differs": {
+   "torch": "2.14.0+cu130 (pinned 2.11.0)",
+   "torchvision": "0.29.1+cu130 (pinned 0.26.0)",
+   "tokenizers": "0.23.3 (pinned 0.23.1)",
+   "numpy": "2.4.6 (pinned 1.26.4)",
+   "pillow": "13.0.0 (pinned 12.2.0)",
+   "accelerate": "1.16.0.dev0 (pinned 1.13.0)",
+   "huggingface-hub": "2.3.0 (pinned 1.32.0)",
+   "sentencepiece": "0.2.2 (pinned 0.2.1)",
+   "protobuf": "7.36.2 (pinned 6.33.6)",
+   "soundfile": "0.14.0 (pinned 0.13.1)",
+   "scipy": "1.17.1 (pinned 1.15.3)"
   },
-  "seconds": 4.43,
   "platform": "cuda"
  },
  "outputs": {
@@ -52,22 +66,22 @@
   ],
   "top5_probs": [
    [
-    0.9963154196739197,
-    0.0035723550245165825,
-    1.4807335446676007e-07,
-    1.479782127944418e-07,
-    1.4519801538881438e-07
+    0.99631267786026,
+    0.0035750409588217735,
+    1.4806560955094028e-07,
+    1.4797329583871033e-07,
+    1.4519096680487564e-07
    ],
    [
     0.9955466985702515,
-    0.00032940020901151,
-    4.835841991734924e-06,
-    4.792936124431435e-06,
-    4.6560203372791875e-06
+    0.00032939735683612525,
+    4.835916115553118e-06,
+    4.793009338754928e-06,
+    4.656082637666259e-06
    ]
   ],
   "training": {
-   "loss": 0.004077321384102106,
+   "loss": 0.00407868716865778,
    "gradients": {
     "vit.embeddings.cls_token": true,
     "vit.embeddings.position_embeddings": true