tarekziade HF Staff
Publish MIC usage, slim fixture and reviewed Transformers expectations
de4a0b8 verified | { | |
| "fixture": "huggingface-mic/switch-base-8-mic", | |
| "key": "c3577706d546056626e717669a48388e9c478d838847e9730c219aa7435641a6", | |
| "key_fields": { | |
| "architectures": [ | |
| "SwitchTransformersForConditionalGeneration" | |
| ], | |
| "output_hidden_states": false, | |
| "return_dict": true, | |
| "chunk_size_feed_forward": 0, | |
| "is_encoder_decoder": true, | |
| "id2label": { | |
| "0": "LABEL_0", | |
| "1": "LABEL_1" | |
| }, | |
| "label2id": { | |
| "LABEL_0": 0, | |
| "LABEL_1": 1 | |
| }, | |
| "problem_type": null, | |
| "vocab_size": 32128, | |
| "d_model": 64, | |
| "d_kv": 16, | |
| "d_ff": 128, | |
| "expert_capacity": 64, | |
| "num_layers": 2, | |
| "num_sparse_encoder_layers": 6, | |
| "num_decoder_layers": 2, | |
| "num_sparse_decoder_layers": 6, | |
| "num_heads": 4, | |
| "num_experts": 4, | |
| "router_bias": false, | |
| "router_jitter_noise": 0, | |
| "router_dtype": "float32", | |
| "router_ignore_padding_tokens": false, | |
| "relative_attention_num_buckets": 32, | |
| "relative_attention_max_distance": 128, | |
| "dropout_rate": 0.1, | |
| "layer_norm_epsilon": 1e-06, | |
| "router_z_loss_coef": 0.001, | |
| "router_aux_loss_coef": 0.001, | |
| "initializer_factor": 1.0, | |
| "dense_act_fn": "relu", | |
| "add_router_probs": false, | |
| "use_cache": true, | |
| "pad_token_id": 0, | |
| "eos_token_id": 1, | |
| "bos_token_id": null, | |
| "tie_word_embeddings": true, | |
| "is_decoder": false, | |
| "add_cross_attention": false, | |
| "encoder_sparse_step": 2, | |
| "decoder_sparse_step": 2, | |
| "batch_prioritized_routing": false, | |
| "feed_forward_proj": "relu", | |
| "decoder_start_token_id": 0, | |
| "is_gated_act": false, | |
| "model_type": "switch_transformers", | |
| "num_selected_experts": 1, | |
| "router_type": "tokens_masked", | |
| "task_specific_params": { | |
| "summarization": { | |
| "early_stopping": true, | |
| "length_penalty": 2.0, | |
| "max_length": 200, | |
| "min_length": 30, | |
| "no_repeat_ngram_size": 3, | |
| "num_beams": 4, | |
| "prefix": "summarize: " | |
| }, | |
| "translation_en_to_de": { | |
| "early_stopping": true, | |
| "max_length": 300, | |
| "num_beams": 4, | |
| "prefix": "translate English to German: " | |
| }, | |
| "translation_en_to_fr": { | |
| "early_stopping": true, | |
| "max_length": 300, | |
| "num_beams": 4, | |
| "prefix": "translate English to French: " | |
| }, | |
| "translation_en_to_ro": { | |
| "early_stopping": true, | |
| "max_length": 300, | |
| "num_beams": 4, | |
| "prefix": "translate English to Romanian: " | |
| } | |
| }, | |
| "output_attentions": false, | |
| "dtype": "bfloat16" | |
| }, | |
| "covers": { | |
| "google/switch-base-8": "92fe2d22b024d9937146fe097ba3d3a7ba146e1b" | |
| }, | |
| "built_from": { | |
| "repo": "google/switch-base-8", | |
| "revision": "92fe2d22b024d9937146fe097ba3d3a7ba146e1b" | |
| }, | |
| "seed": 17, | |
| "dtype": "bfloat16", | |
| "builder": "example/build_fixtures.py", | |
| "builder_revision": "1beded04f8b71c7209928f32f6d28099aea4accc", | |
| "recipe": { | |
| "seed": 17, | |
| "model_class": "SwitchTransformersForConditionalGeneration" | |
| }, | |
| "shrink": "rule", | |
| "model_class": "SwitchTransformersForConditionalGeneration", | |
| "parameters": 2319872, | |
| "missing_parameters": [], | |
| "serialization": { | |
| "file": "model.safetensors", | |
| "source_file": "pytorch_model.bin", | |
| "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": "SwitchTransformersForConditionalGeneration" | |
| }, | |
| "weights": "random", | |
| "gate": { | |
| "random": { | |
| "pass": true, | |
| "accepted_insensitive": {}, | |
| "threshold": 0.001, | |
| "deltas": { | |
| "input_ids:roll_batch": 4.765625, | |
| "input_ids:reverse": 4.8125, | |
| "attention_mask:all_ones": 4.359375 | |
| }, | |
| "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", | |
| "spiece.model", | |
| "tokenizer.json", | |
| "tokenizer_config.json" | |
| ], | |
| "environment": { | |
| "python": "3.11.18", | |
| "machine": "x86_64", | |
| "torch": "2.15.0+cpu", | |
| "transformers": "5.23.0.dev0", | |
| "tokenizers": "0.23.3" | |
| }, | |
| "sha256": { | |
| "config.json": "6e5f33eed467fcbffceb24429acb12b8a4d7a5583a9f81a5cd311299ba45394a", | |
| "generation_config.json": "f5a1c7e2be8092018d8835128987edf0111637dd98e90599cc80310fef75d95a", | |
| "model.safetensors": "78add58bb9b644e98a44d253077d6275acf2700abfae5833e468f3c38e38af52", | |
| "special_tokens_map.json": "5c87151ef0f72a99d1f766a4c418bd2a1f90aaa30a8e22fe5eca9641daebb64f", | |
| "spiece.model": "d60acb128cf7b7f2536e8f38a5b18a05535c9e14c7a355904270e15b0945ea86", | |
| "tokenizer.json": "5f0ed8ab5b8cfa9812bb73752f1d80c292e52bcf5a87a144dc9ab2d251056cbb", | |
| "tokenizer_config.json": "4969f8d76ef05a16553bd2b07b3501673ae8d36972aea88a0f78ad31a3ff2de9" | |
| } | |
| } | |