| { |
| "architectures": [ |
| "MicroDistilleryBigramForCausalLM" |
| ], |
| "model_type": "custom", |
| "custom_model_type": "micro-distillery-bigram", |
| "vocab_size": 27, |
| "max_position_embeddings": 1024, |
| "hidden_size": 512, |
| "num_hidden_layers": 0, |
| "num_attention_heads": 0, |
| "activation_function": "gelu_new", |
| "bos_token_id": null, |
| "eos_token_id": null, |
| "is_toy_model": true, |
| "training_method": "simplified-policy-gradient", |
| "vae_note": "VAE latent_dim and filter_threshold are recorded for reference only; they do not affect the transition table.", |
| "eos_note": "No explicit EOS token is emitted; generation stops at max_tokens. BOS (space) is the practical trajectory EOS signal. bos_token_id/eos_token_id are null because no such ids exist in vocab.json.", |
| "micro_distillery": { |
| "format": "mD-v2", |
| "version": "2.5.0", |
| "harmonic_seed": 1251963776, |
| "grpo_config": { |
| "group_size": 8, |
| "kl_penalty": 0.1, |
| "training_steps": 60 |
| }, |
| "vae_config": { |
| "latent_dim": 32, |
| "filter_threshold": 0.7, |
| "applied": false |
| } |
| } |
| } |
