| { | |
| "name": "pixel_100m", | |
| "vocab_size": 1262, | |
| "context_length": 1024, | |
| "num_layers": 12, | |
| "hidden_size": 768, | |
| "num_attention_heads": 12, | |
| "num_key_value_heads": 4, | |
| "intermediate_size": 2048, | |
| "rope_base": 500000, | |
| "tie_word_embeddings": true, | |
| "dropout": 0.0, | |
| "attention_dropout": 0.0, | |
| "rms_norm_eps": 1e-05, | |
| "use_moe": false, | |
| "moe": { | |
| "enabled": false, | |
| "num_experts": 4, | |
| "top_k": 2, | |
| "expert_interval": 4 | |
| }, | |
| "lora": { | |
| "enabled": false, | |
| "rank": 8, | |
| "alpha": 16, | |
| "dropout": 0.05, | |
| "target_modules": [ | |
| "q_proj", | |
| "k_proj", | |
| "v_proj", | |
| "o_proj", | |
| "gate_proj", | |
| "up_proj", | |
| "down_proj" | |
| ] | |
| } | |
| } |