| { | |
| "model_id": "datalab-to/surya-ocr-2", | |
| "mode": "decode_step", | |
| "source_dtype": "bf16", | |
| "coreml_compute_precision": "fp16", | |
| "max_cache_length": 512, | |
| "prompt_tokens_for_trace": 300, | |
| "package": "/Users/studio/datalab-quants-cairo/artifacts/coreml/surya-ocr-2-coreml-8bit/surya_decode_step_fp16_cache512.mlpackage", | |
| "inputs": [ | |
| "inputs_embeds", | |
| "cos", | |
| "sin", | |
| "attention_mask", | |
| "full_key_0", | |
| "full_key_1", | |
| "full_key_2", | |
| "full_key_3", | |
| "full_key_4", | |
| "full_key_5", | |
| "full_value_0", | |
| "full_value_1", | |
| "full_value_2", | |
| "full_value_3", | |
| "full_value_4", | |
| "full_value_5", | |
| "conv_state_0", | |
| "conv_state_1", | |
| "conv_state_2", | |
| "conv_state_3", | |
| "conv_state_4", | |
| "conv_state_5", | |
| "conv_state_6", | |
| "conv_state_7", | |
| "conv_state_8", | |
| "conv_state_9", | |
| "conv_state_10", | |
| "conv_state_11", | |
| "conv_state_12", | |
| "conv_state_13", | |
| "conv_state_14", | |
| "conv_state_15", | |
| "conv_state_16", | |
| "conv_state_17", | |
| "recurrent_state_0", | |
| "recurrent_state_1", | |
| "recurrent_state_2", | |
| "recurrent_state_3", | |
| "recurrent_state_4", | |
| "recurrent_state_5", | |
| "recurrent_state_6", | |
| "recurrent_state_7", | |
| "recurrent_state_8", | |
| "recurrent_state_9", | |
| "recurrent_state_10", | |
| "recurrent_state_11", | |
| "recurrent_state_12", | |
| "recurrent_state_13", | |
| "recurrent_state_14", | |
| "recurrent_state_15", | |
| "recurrent_state_16", | |
| "recurrent_state_17" | |
| ], | |
| "outputs": [ | |
| "logits", | |
| "new_full_key_0", | |
| "new_full_key_1", | |
| "new_full_key_2", | |
| "new_full_key_3", | |
| "new_full_key_4", | |
| "new_full_key_5", | |
| "new_full_value_0", | |
| "new_full_value_1", | |
| "new_full_value_2", | |
| "new_full_value_3", | |
| "new_full_value_4", | |
| "new_full_value_5", | |
| "new_conv_state_0", | |
| "new_conv_state_1", | |
| "new_conv_state_2", | |
| "new_conv_state_3", | |
| "new_conv_state_4", | |
| "new_conv_state_5", | |
| "new_conv_state_6", | |
| "new_conv_state_7", | |
| "new_conv_state_8", | |
| "new_conv_state_9", | |
| "new_conv_state_10", | |
| "new_conv_state_11", | |
| "new_conv_state_12", | |
| "new_conv_state_13", | |
| "new_conv_state_14", | |
| "new_conv_state_15", | |
| "new_conv_state_16", | |
| "new_conv_state_17", | |
| "new_recurrent_state_0", | |
| "new_recurrent_state_1", | |
| "new_recurrent_state_2", | |
| "new_recurrent_state_3", | |
| "new_recurrent_state_4", | |
| "new_recurrent_state_5", | |
| "new_recurrent_state_6", | |
| "new_recurrent_state_7", | |
| "new_recurrent_state_8", | |
| "new_recurrent_state_9", | |
| "new_recurrent_state_10", | |
| "new_recurrent_state_11", | |
| "new_recurrent_state_12", | |
| "new_recurrent_state_13", | |
| "new_recurrent_state_14", | |
| "new_recurrent_state_15", | |
| "new_recurrent_state_16", | |
| "new_recurrent_state_17" | |
| ], | |
| "state_contract": { | |
| "full_attention_layers": 6, | |
| "linear_attention_layers": 18, | |
| "host_updates_full_kv_cache": true, | |
| "host_updates_token_position_and_attention_mask": true | |
| } | |
| } | |