tarekziade HF Staff
Consolidate MIC usage, slim weights and reviewed framework expectations
8d226a7 verified | { | |
| "fixture": "hf-internal-testing/mic-llama-slim", | |
| "key": "449c36ea4c20ef20fee1bc0f6e79e16d73feeb50f80b4a3a7eac445b2cb36258", | |
| "key_fields": { | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "output_hidden_states": false, | |
| "return_dict": true, | |
| "chunk_size_feed_forward": 0, | |
| "is_encoder_decoder": false, | |
| "id2label": { | |
| "0": "LABEL_0", | |
| "1": "LABEL_1" | |
| }, | |
| "label2id": { | |
| "LABEL_0": 0, | |
| "LABEL_1": 1 | |
| }, | |
| "problem_type": null, | |
| "vocab_size": 49152, | |
| "hidden_size": 64, | |
| "intermediate_size": 128, | |
| "num_hidden_layers": 2, | |
| "num_attention_heads": 4, | |
| "num_key_value_heads": 2, | |
| "hidden_act": "silu", | |
| "max_position_embeddings": 8192, | |
| "initializer_range": 0.041666666666666664, | |
| "rms_norm_eps": 1e-05, | |
| "use_cache": true, | |
| "pad_token_id": null, | |
| "bos_token_id": 0, | |
| "eos_token_id": 0, | |
| "pretraining_tp": 1, | |
| "tie_word_embeddings": true, | |
| "rope_parameters": { | |
| "rope_theta": 100000, | |
| "rope_type": "default" | |
| }, | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "mlp_bias": false, | |
| "head_dim": 16, | |
| "is_llama_config": true, | |
| "model_type": "llama", | |
| "rope_interleaved": false, | |
| "output_attentions": false, | |
| "dtype": "bfloat16" | |
| }, | |
| "covers": { | |
| "HuggingFaceTB/SmolLM2-135M": "93efa2f097d58c2a74874c7e644dbc9b0cee75a2" | |
| }, | |
| "built_from": { | |
| "repo": "HuggingFaceTB/SmolLM2-135M", | |
| "revision": "93efa2f097d58c2a74874c7e644dbc9b0cee75a2" | |
| }, | |
| "seed": 17, | |
| "dtype": "bfloat16", | |
| "builder": "example/build_fixtures.py", | |
| "model_class": "LlamaForCausalLM", | |
| "parameters": 3219776, | |
| "missing_parameters": [], | |
| "checkpoint_inspection": "parameter metadata", | |
| "captured": { | |
| "call": "training step", | |
| "inputs": [ | |
| "attention_mask", | |
| "input_ids", | |
| "labels" | |
| ] | |
| }, | |
| "weights": "random", | |
| "gate": { | |
| "random": { | |
| "pass": true, | |
| "accepted_insensitive": {}, | |
| "threshold": 0.001, | |
| "deltas": { | |
| "input_ids:roll_batch": 2.5703125, | |
| "input_ids:reverse": 2.1171875, | |
| "attention_mask:all_ones": null | |
| }, | |
| "weakest": "input_ids:reverse", | |
| "no_perturbation_defined": [] | |
| } | |
| }, | |
| "numeric_health": { | |
| "float32": { | |
| "outputs_finite": true, | |
| "loss_finite": true, | |
| "gradients_finite": true, | |
| "finite": true | |
| }, | |
| "bfloat16": { | |
| "outputs_finite": true, | |
| "loss_finite": true, | |
| "gradients_finite": true, | |
| "finite": true | |
| }, | |
| "float16": { | |
| "outputs_finite": true, | |
| "loss_finite": true, | |
| "gradients_finite": true, | |
| "finite": true | |
| } | |
| }, | |
| "copied_unchanged": [ | |
| "generation_config.json", | |
| "merges.txt", | |
| "special_tokens_map.json", | |
| "tokenizer.json", | |
| "tokenizer_config.json", | |
| "vocab.json" | |
| ], | |
| "environment": { | |
| "python": "3.10.12", | |
| "machine": "x86_64", | |
| "torch": "2.11.0+cu128", | |
| "transformers": "5.19.0.dev0", | |
| "tokenizers": "0.23.1" | |
| }, | |
| "sha256": { | |
| "config.json": "5b67e6e85527da68b95f91bb2880a703c69088209663f4c5fd5ee2b03a4df9d7", | |
| "generation_config.json": "2056c988e990b0d13670f63f2f3b87b3b6d07edaf7a3416998ba27dab2d8a059", | |
| "merges.txt": "0b54e8aa4e53d5383e2e4bc635a56b43f9647f7b13832d5d9ecd8f82dac4f510", | |
| "model.safetensors": "7b6d8c2bf680c2a12efbf241245aedda9ebdd7ecae028bc768ca2ca925c054ef", | |
| "special_tokens_map.json": "e786b595b9a23148bf1630df78d9037a048ea671e48bfd3549a1e3c233742bb3", | |
| "tokenizer.json": "9ca9acddb6525a194ec8ac7a87f24fbba7232a9a15ffa1af0c1224fcd888e47c", | |
| "tokenizer_config.json": "4bb9af56a342753d39374f4016a16574cab299fe088e896f425ce3c433f61424", | |
| "vocab.json": "82b84012e3add4d01d12ba14442026e49b8cbbaead1f79ecf3d919784f82dc79" | |
| } | |
| } | |