| { |
| "architectures": [ |
| "ModernBertDecisionModel" |
| ], |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "bos_token_id": 2, |
| "classifier_activation": "gelu", |
| "classifier_bias": false, |
| "classifier_dropout": 0.0, |
| "classifier_pooling": "mean", |
| "cls_token_id": 1, |
| "decoder_bias": true, |
| "deterministic_flash_attn": false, |
| "dtype": "float32", |
| "embedding_dropout": 0.0, |
| "eos_token_id": 1, |
| "global_attn_every_n_layers": 3, |
| "gradient_checkpointing": false, |
| "hidden_activation": "gelu", |
| "hidden_size": 768, |
| "initializer_cutoff_factor": 2.0, |
| "initializer_range": 0.02, |
| "intermediate_size": 1152, |
| "layer_norm_eps": 1e-05, |
| "layer_types": [ |
| "full_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention" |
| ], |
| "local_attention": 128, |
| "mask_token_id": 4, |
| "max_position_embeddings": 8192, |
| "mlp_bias": false, |
| "mlp_dropout": 0.0, |
| "model_type": "modernbert", |
| "norm_bias": false, |
| "norm_eps": 1e-05, |
| "num_attention_heads": 12, |
| "num_hidden_layers": 22, |
| "pad_token_id": 0, |
| "position_embedding_type": "sans_pos", |
| "repad_logits_with_grad": false, |
| "rope_parameters": { |
| "full_attention": { |
| "rope_theta": 160000, |
| "rope_type": "default" |
| }, |
| "sliding_attention": { |
| "rope_theta": 160000, |
| "rope_type": "default" |
| } |
| }, |
| "sep_token_id": 1, |
| "sparse_pred_ignore_index": -100, |
| "sparse_prediction": false, |
| "tie_word_embeddings": true, |
| "transformers_version": "5.0.0", |
| "vocab_size": 256000, |
| "decision": { |
| "encoder": "jhu-clsp/mmBERT-base", |
| "head_layers": 2, |
| "max_len": 1024, |
| "head_max_len": 512, |
| "max_prefixes": 6, |
| "act_costs": { |
| "escalate": 0.5 |
| }, |
| "cost_wrong_act": 3.0, |
| "amp_dtype": "bf16", |
| "model_name": "GoMini", |
| "temperature": [ |
| 1.0, |
| 1.0, |
| 1.0 |
| ], |
| "temperature_by_options": {}, |
| "auto_head_budget": false, |
| "tournament_above": 24, |
| "training": { |
| "objective": "RLCD — strictly proper scoring reward (log + spherical + ranked probability score)", |
| "epochs": 2, |
| "steps": 5468, |
| "items": 157912, |
| "languages": "az, en, ru, tr, de, fr, es, ar, hi, zh, fa, ka, ja, ko, uk, kk, uz, it", |
| "held_out_for_evaluation": "SIB-200, AG News, DAIR Emotion, Banking77, SMS spam, hand-written routing set, 5% of synthetic schemas", |
| "post_training": { |
| "stage_2": "spatial-control (procedural camera/cursor/drone/robot/car tasks, no game data), 843 steps", |
| "stage_3": "LLM/agent-routing emphasis with replay, 770 steps", |
| "stage_4": "generic skills: bare-key options, exact identifier lookup, no-good-option honesty (procedural + transformed training data)", |
| "stage_5": "v7-v9 robustness (az/ru/en): ~600 new business taxonomies incl. hard confusable pairs, clean+messy messages (Wahoo, GPT-6-Luna, Claude Sonnet 5.5), programmatic translit / no-diacritics / typo corruptions, 15-category support-ticket taxonomy with boundary cases; weight soup 0.85 v9 + 0.15 v6", |
| "stage_6": "native number questions: value lookup among stated numbers + explicit not-stated option (15.5k Wahoo-verified az/ru/en items) with full replay; weight soup 0.85 number stage + 0.15 release3", |
| "stage_7": "v10: number lookup (native + range buckets), excerpt (part + word, start/end; document chunks), citation; 29k number and 14k excerpt questions verified (Wahoo 1.5, GPT-6-Luna, Claude Sonnet 5.5) + 11k new hard decision messages, full replay; continuation with distractor-aware range buckets; soup 0.85 v10 + 0.15 release4; v10c: currency equivalence ($ = US dollars; other currencies -> not stated)" |
| } |
| }, |
| "gomini": { |
| "schema": "gomini.agentgo/1", |
| "training": "Full encoder, embeddings, typed decision head; no adapters", |
| "optimizer_steps": 5660, |
| "auxiliary_act_head": "unsupervised, excluded from execution and optimization", |
| "confidence": "entropy_confidence recomputed by common adapter; never sufficient for permission", |
| "calibration": "required separately" |
| } |
| } |
| } |
|
|
