File size: 4,150 Bytes
ad271a3 d52a4cb ad271a3 d52a4cb ad271a3 d52a4cb ad271a3 d52a4cb ad271a3 ea5b7a6 ad271a3 d52a4cb ad271a3 | 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 | {
"mic": "0.2",
"name": "SmolLM text generation",
"usage": "AutoTokenizer + AutoModelForCausalLM.generate, batched, left padding",
"architecture": "LlamaForCausalLM",
"checkpoints": {
"HuggingFaceTB/SmolLM-135M": {
"fixture": "hf-internal-testing/mic-SmolLM-slim"
},
"HuggingFaceTB/SmolLM-360M": {
"fixture": "hf-internal-testing/mic-SmolLM-slim"
},
"HuggingFaceTB/SmolLM-1.7B": {
"fixture": "hf-internal-testing/mic-SmolLM-1.7B-slim"
}
},
"fixtures": {
"hf-internal-testing/mic-SmolLM-slim": {
"built_from": "HuggingFaceTB/SmolLM-135M",
"covers": [
"HuggingFaceTB/SmolLM-135M",
"HuggingFaceTB/SmolLM-360M"
],
"build": {
"seed": 17,
"overfit": "random weights generate the same tokens for every prompt, so generated_tokens cannot tell the prompts apart. The training rows share 'The capital of' and then name four countries, so ln 4 at that position (about 0.2 per token) cannot be learned: target_loss 0.3",
"target_loss": 0.3
}
},
"hf-internal-testing/mic-SmolLM-1.7B-slim": {
"built_from": "HuggingFaceTB/SmolLM-1.7B",
"covers": [
"HuggingFaceTB/SmolLM-1.7B"
],
"why_separate": "different tokenizer.json and tokenizer_config.json; full multi-head attention instead of grouped key/value heads"
}
},
"integrations": {
"transformers": {
"entry_point": "integrations/transformers/usage.md",
"format": "runnable-markdown",
"requires": [
"transformers>=5",
"torch",
"accelerate"
],
"targets": [
"fixtures",
"checkpoints"
],
"variants": {
"default": {},
"eager": {
"attn_implementation": "eager"
},
"bfloat16": {
"dtype": "bfloat16"
}
},
"training": true,
"resources": {
"fixture": {
"memory_gb": 1,
"download_mb": 16
},
"HuggingFaceTB/SmolLM-135M": {
"memory_gb": 2,
"download_mb": 513
},
"HuggingFaceTB/SmolLM-360M": {
"memory_gb": 3,
"download_mb": 1300
},
"HuggingFaceTB/SmolLM-1.7B": {
"memory_gb": 12,
"disk_gb": 8,
"download_mb": 6400,
"measured": "peak RSS 10.6 GB (default) and 11.3 GB (bfloat16: the variant reload holds two copies), one process, Apple M5, 2026-10-07"
}
}
},
"transformers.js": {
"entry_point": "integrations/transformers.js/integration.js",
"format": "script",
"requires": [
"@huggingface/transformers@4.3.0"
],
"targets": [
"checkpoints"
],
"artifacts": {
"HuggingFaceTB/SmolLM-135M": {
"repo": "onnx-community/SmolLM-135M-ONNX",
"file": "onnx/model.onnx"
},
"HuggingFaceTB/SmolLM-360M": {
"repo": "onnx-community/SmolLM-360M-ONNX",
"file": "onnx/model.onnx"
}
},
"unsupported": {
"fixtures": "no ONNX conversion of the fixtures yet",
"HuggingFaceTB/SmolLM-1.7B": "no onnx-community conversion; the creator's float32 ONNX is 6.8 GB"
},
"variants": {
"default": {
"dtype": "fp32"
},
"webgpu": {
"dtype": "fp32",
"device": "webgpu"
},
"webgpu-fp16": {
"dtype": "fp16",
"device": "webgpu"
},
"webgpu-q4f16": {
"dtype": "q4f16",
"device": "webgpu"
}
},
"training": false
}
},
"outputs": {
"runtime": {
"compare": "exact"
},
"generated_tokens": {
"compare": "exact"
},
"generated_text": {
"compare": "exact"
},
"next_token_logits": {
"compare": "allclose",
"atol": 0.0001,
"rtol": 0.0001
},
"training.loss": {
"compare": "allclose",
"atol": 0.0001,
"rtol": 0.0001
},
"training.gradients": {
"compare": "all_true"
},
"training.updated": {
"compare": "true"
}
}
}
|