tarekziade HF Staff
Publish MIC usage, slim fixture and reviewed Transformers expectations
efb1356 verified | { | |
| "fixture": "huggingface-mic/tiny-aya-global-mic", | |
| "key": "7ec67d2caa1674bf17f5a05c3a83132c771e55344c411d9b1dc8208b35ee2003", | |
| "key_fields": { | |
| "architectures": [ | |
| "Cohere2ForCausalLM" | |
| ], | |
| "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": 262144, | |
| "hidden_size": 64, | |
| "intermediate_size": 128, | |
| "logit_scale": 1.0, | |
| "num_hidden_layers": 2, | |
| "num_attention_heads": 4, | |
| "num_key_value_heads": 2, | |
| "hidden_act": "silu", | |
| "max_position_embeddings": 8192, | |
| "initializer_range": 0.02, | |
| "layer_norm_eps": 1e-05, | |
| "use_cache": true, | |
| "pad_token_id": 0, | |
| "bos_token_id": 2, | |
| "eos_token_id": 3, | |
| "tie_word_embeddings": true, | |
| "rope_parameters": { | |
| "rope_theta": 50000, | |
| "rope_type": "default" | |
| }, | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "sliding_window": 4096, | |
| "layer_types": [ | |
| "sliding_attention", | |
| "full_attention" | |
| ], | |
| "head_dim": 16, | |
| "_sliding_window_pattern": 4, | |
| "cache_implementation": "hybrid", | |
| "layer_switch": 4, | |
| "model_type": "cohere2", | |
| "order_of_interleaved_layers": "local_attn_first", | |
| "position_embedding_type": "rope_gptj", | |
| "rotary_pct": 1.0, | |
| "sliding_window_pattern": 4, | |
| "use_embedding_sharing": true, | |
| "use_gated_activation": true, | |
| "use_parallel_block": true, | |
| "use_parallel_embedding": false, | |
| "use_qk_norm": false, | |
| "output_attentions": false, | |
| "dtype": "bfloat16" | |
| }, | |
| "covers": { | |
| "CohereLabs/tiny-aya-global": "af89d219b53ed9b13b8a4645f9c8028973510324" | |
| }, | |
| "built_from": { | |
| "repo": "CohereLabs/tiny-aya-global", | |
| "revision": "af89d219b53ed9b13b8a4645f9c8028973510324" | |
| }, | |
| "seed": 17, | |
| "dtype": "bfloat16", | |
| "builder": "example/build_fixtures.py", | |
| "builder_revision": "4fe1f16c8fa5ca6e12a45df69cc19e5c6791e39c", | |
| "recipe": { | |
| "seed": 17, | |
| "model_class": "Cohere2ForCausalLM" | |
| }, | |
| "shrink": "rule", | |
| "model_class": "Cohere2ForCausalLM", | |
| "parameters": 16851136, | |
| "missing_parameters": [], | |
| "serialization": { | |
| "file": "model.safetensors", | |
| "reverted_conversions": [] | |
| }, | |
| "checkpoint_inspection": "parameter headers/indexes; bounded binary fallback", | |
| "numeric_health_scope": "CPU float32/bfloat16/float16; autocast for low-precision losses", | |
| "captured": { | |
| "call": "training step", | |
| "inputs": [ | |
| "attention_mask", | |
| "input_ids", | |
| "labels" | |
| ], | |
| "class": "Cohere2ForCausalLM" | |
| }, | |
| "weights": "random", | |
| "gate": { | |
| "random": { | |
| "pass": true, | |
| "accepted_insensitive": {}, | |
| "threshold": 0.001, | |
| "deltas": { | |
| "input_ids:roll_batch": 1.31591796875, | |
| "input_ids:reverse": 1.46875, | |
| "attention_mask:all_ones": 0.8212890625 | |
| }, | |
| "weakest": "attention_mask:all_ones", | |
| "no_perturbation_defined": [], | |
| "not_a_forward_parameter": [] | |
| } | |
| }, | |
| "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, | |
| "unscaled_gradients_finite": true, | |
| "finite": true | |
| } | |
| }, | |
| "copied_unchanged": [ | |
| "generation_config.json", | |
| "special_tokens_map.json", | |
| "tokenizer.json", | |
| "tokenizer_config.json" | |
| ], | |
| "environment": { | |
| "python": "3.12.12", | |
| "machine": "arm64", | |
| "torch": "2.14.1", | |
| "transformers": "5.22.0.dev0", | |
| "tokenizers": "0.23.3" | |
| }, | |
| "sha256": { | |
| "config.json": "2babad502695d980ffe2222cce28cefb1418ca07747081f041ae548c519447e2", | |
| "generation_config.json": "0efa696e1ebc0484f8e16fcd1141dba1e897c4699ff971052e43ce5d6d4bf4d0", | |
| "model.safetensors": "7468ee6ee93f0902ff1445d207d3280a175dac3d3b0f67256969e65c9a7e4740", | |
| "special_tokens_map.json": "2b7ce39630f8a18705331a9b32c3b2236e96e7410791f98a9f8b6494e3ab2ea6", | |
| "tokenizer.json": "2227ea9c52e8afb3f98bfed2679008b275f2664de69dfde174b374389eb0225d", | |
| "tokenizer_config.json": "fb56a3ef7659346aae196a2ab877355f1e7fd871c36843f709bb2e23c606d972" | |
| } | |
| } | |