| { |
| "base_model": "/home/ggb/mergehell/outputs/k3_fp8_experts", |
| "common_dir": "/home/ggb/mergehell/outputs/k3_headstack_rowwise_common", |
| "common_file": "model-00102-common-non-o.safetensors", |
| "config_updates": { |
| "n_shared_experts": 2, |
| "num_attention_heads": 128, |
| "num_key_value_heads": 128 |
| }, |
| "created_utc": "2026-06-13T20:17:22Z", |
| "expert_shards": { |
| "count": 102, |
| "mode": "relative_symlink", |
| "source": "/home/ggb/mergehell/outputs/k3_fp8_experts" |
| }, |
| "logical_total_size": 3513903759664, |
| "logical_total_size_gib": 3272.577896401286, |
| "o_proj_file": "model-00103-o-proj-scale050.safetensors", |
| "o_proj_scale": 0.5, |
| "o_proj_total_size": 24192745472, |
| "o_proj_total_size_gib": 22.53125, |
| "output_dir": "/home/ggb/mergehell/outputs/k3_headstack_rowwise_o_050", |
| "rowwise_a_proj_keys": [ |
| "model.layers.N.self_attn.q_a_proj.weight", |
| "model.layers.N.self_attn.kv_a_proj_with_mqa.weight" |
| ], |
| "rowwise_a_proj_merge": true, |
| "shared_output_scale": 0.5, |
| "slerp_dot_threshold": 0.9995, |
| "slerp_eps": 1e-08, |
| "slerp_t": 0.5, |
| "tensor_count": 471562 |
| } |
|
|