checkpoints/gptoss_phase1_config.json

{
  "variant": "gptoss_phase1",
  "model_type": "adamba-moe",
  "architecture": "HybridMoEGPT (Attention + MoE + Mamba)",
  "base_model": "gpt-oss-20b",
  "parameters": "21.9B",
  "n_embd": 2880,
  "features": [
    "mamba_integration",
    "moe_32experts"
  ],
  "n_layers": 36,
  "vocab_size": 201088,
  "num_experts": 32,
  "experts_per_token": 4
}