| { |
| "format": 1, |
| "name": "ridgefill-projector-qwen3.8-flash-next-i8", |
| "authors": [ |
| "Dylan Johnston", |
| "tcclaviger" |
| ], |
| "license": "Apache-2.0", |
| "doi": "10.5281/zenodo.23179168", |
| "homepage": "https://huggingface.co/Dyluhn/ridgefill-projector-qwen3.8-flash-next-i8", |
| "plugin_min_version": "0.3.0", |
| "adapter": "qwen4exp", |
| "adapter_abi": 1, |
| "model": { |
| "arch_id": "qwen4exp", |
| "name": "qwen4exp", |
| "meta": { |
| "model_type": "qwen4_exp", |
| "n_layers": "48", |
| "n_embd": "2560", |
| "n_vocab": "248320", |
| "n_expert": "512", |
| "hc_count": "4", |
| "layer_types": "linear_attention linear_attention linear_attention full_attention linear_attention linear_attention linear_attention full_attention linear_attention linear_attention linear_attention full_attention linear_attention linear_attention linear_attention full_attention linear_attention linear_attention linear_attention full_attention linear_attention linear_attention linear_attention full_attention linear_attention linear_attention linear_attention full_attention linear_attention linear_attention linear_attention full_attention linear_attention linear_attention linear_attention full_attention linear_attention linear_attention linear_attention full_attention linear_attention linear_attention linear_attention full_attention linear_attention linear_attention linear_attention full_attention", |
| "full_attention_interval": "4", |
| "linear_num_value_heads": "48", |
| "linear_value_head_dim": "128", |
| "linear_key_head_dim": "128" |
| }, |
| "vocab_sha256": "17f26a1adbf691d98d9dc9e03237c7ce58b52c28e1e79be46a07899d11455b1d", |
| "encodings": { |
| "blk.23.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.23.attn_hc_inject.weight": "bf16", |
| "blk.23.attn_hc_norm.weight": "bf16", |
| "blk.23.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.23.attn_k.weight": "i8*bf16[1x128]", |
| "blk.23.attn_k_norm.weight": "bf16", |
| "blk.23.attn_output.weight": "i8*bf16[1x128]", |
| "blk.23.attn_q_norm.weight": "bf16", |
| "blk.23.attn_qg.weight": "i8*bf16[1x128]", |
| "blk.23.attn_v.weight": "i8*bf16[1x128]", |
| "blk.23.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.23.ffn_gate_inp.weight": "bf16", |
| "blk.23.ffn_gate_shexp.weight": "bf16", |
| "blk.23.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.23.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.23.ffn_hc_inject.weight": "bf16", |
| "blk.23.ffn_hc_norm.weight": "bf16", |
| "blk.23.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.23.qsa_k_norm.weight": "bf16", |
| "blk.23.qsa_q_norm.weight": "bf16", |
| "blk.23.qsa_qk.weight": "bf16", |
| "blk.24.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.24.attn_hc_inject.weight": "bf16", |
| "blk.24.attn_hc_norm.weight": "bf16", |
| "blk.24.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.24.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.24.ffn_gate_inp.weight": "bf16", |
| "blk.24.ffn_gate_shexp.weight": "bf16", |
| "blk.24.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.24.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.24.ffn_hc_inject.weight": "bf16", |
| "blk.24.ffn_hc_norm.weight": "bf16", |
| "blk.24.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.24.ssm_a_log": "bf16", |
| "blk.24.ssm_ab.weight": "bf16", |
| "blk.24.ssm_conv1d.weight": "bf16", |
| "blk.24.ssm_dt.bias": "bf16", |
| "blk.24.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.24.ssm_norm.weight": "bf16", |
| "blk.24.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.25.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.25.attn_hc_inject.weight": "bf16", |
| "blk.25.attn_hc_norm.weight": "bf16", |
| "blk.25.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.25.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.25.ffn_gate_inp.weight": "bf16", |
| "blk.25.ffn_gate_shexp.weight": "bf16", |
| "blk.25.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.25.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.25.ffn_hc_inject.weight": "bf16", |
| "blk.25.ffn_hc_norm.weight": "bf16", |
| "blk.25.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.25.ssm_a_log": "bf16", |
| "blk.25.ssm_ab.weight": "bf16", |
| "blk.25.ssm_conv1d.weight": "bf16", |
| "blk.25.ssm_dt.bias": "bf16", |
| "blk.25.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.25.ssm_norm.weight": "bf16", |
| "blk.25.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.26.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.26.attn_hc_inject.weight": "bf16", |
| "blk.26.attn_hc_norm.weight": "bf16", |
| "blk.26.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.26.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.26.ffn_gate_inp.weight": "bf16", |
| "blk.26.ffn_gate_shexp.weight": "bf16", |
| "blk.26.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.26.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.26.ffn_hc_inject.weight": "bf16", |
| "blk.26.ffn_hc_norm.weight": "bf16", |
| "blk.26.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.26.ssm_a_log": "bf16", |
| "blk.26.ssm_ab.weight": "bf16", |
| "blk.26.ssm_conv1d.weight": "bf16", |
| "blk.26.ssm_dt.bias": "bf16", |
| "blk.26.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.26.ssm_norm.weight": "bf16", |
| "blk.26.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.27.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.27.attn_hc_inject.weight": "bf16", |
| "blk.27.attn_hc_norm.weight": "bf16", |
| "blk.27.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.27.attn_k.weight": "i8*bf16[1x128]", |
| "blk.27.attn_k_norm.weight": "bf16", |
| "blk.27.attn_output.weight": "i8*bf16[1x128]", |
| "blk.27.attn_q_norm.weight": "bf16", |
| "blk.27.attn_qg.weight": "i8*bf16[1x128]", |
| "blk.27.attn_v.weight": "i8*bf16[1x128]", |
| "blk.27.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.27.ffn_gate_inp.weight": "bf16", |
| "blk.27.ffn_gate_shexp.weight": "bf16", |
| "blk.27.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.27.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.27.ffn_hc_inject.weight": "bf16", |
| "blk.27.ffn_hc_norm.weight": "bf16", |
| "blk.27.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.27.qsa_k_norm.weight": "bf16", |
| "blk.27.qsa_q_norm.weight": "bf16", |
| "blk.27.qsa_qk.weight": "bf16", |
| "blk.28.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.28.attn_hc_inject.weight": "bf16", |
| "blk.28.attn_hc_norm.weight": "bf16", |
| "blk.28.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.28.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.28.ffn_gate_inp.weight": "bf16", |
| "blk.28.ffn_gate_shexp.weight": "bf16", |
| "blk.28.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.28.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.28.ffn_hc_inject.weight": "bf16", |
| "blk.28.ffn_hc_norm.weight": "bf16", |
| "blk.28.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.28.ssm_a_log": "bf16", |
| "blk.28.ssm_ab.weight": "bf16", |
| "blk.28.ssm_conv1d.weight": "bf16", |
| "blk.28.ssm_dt.bias": "bf16", |
| "blk.28.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.28.ssm_norm.weight": "bf16", |
| "blk.28.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.29.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.29.attn_hc_inject.weight": "bf16", |
| "blk.29.attn_hc_norm.weight": "bf16", |
| "blk.29.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.29.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.29.ffn_gate_inp.weight": "bf16", |
| "blk.29.ffn_gate_shexp.weight": "bf16", |
| "blk.29.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.29.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.29.ffn_hc_inject.weight": "bf16", |
| "blk.29.ffn_hc_norm.weight": "bf16", |
| "blk.29.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.29.ssm_a_log": "bf16", |
| "blk.29.ssm_ab.weight": "bf16", |
| "blk.29.ssm_conv1d.weight": "bf16", |
| "blk.29.ssm_dt.bias": "bf16", |
| "blk.29.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.29.ssm_norm.weight": "bf16", |
| "blk.29.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.30.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.30.attn_hc_inject.weight": "bf16", |
| "blk.30.attn_hc_norm.weight": "bf16", |
| "blk.30.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.30.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.30.ffn_gate_inp.weight": "bf16", |
| "blk.30.ffn_gate_shexp.weight": "bf16", |
| "blk.30.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.30.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.30.ffn_hc_inject.weight": "bf16", |
| "blk.30.ffn_hc_norm.weight": "bf16", |
| "blk.30.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.30.ssm_a_log": "bf16", |
| "blk.30.ssm_ab.weight": "bf16", |
| "blk.30.ssm_conv1d.weight": "bf16", |
| "blk.30.ssm_dt.bias": "bf16", |
| "blk.30.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.30.ssm_norm.weight": "bf16", |
| "blk.30.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.31.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.31.attn_hc_inject.weight": "bf16", |
| "blk.31.attn_hc_norm.weight": "bf16", |
| "blk.31.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.31.attn_k.weight": "i8*bf16[1x128]", |
| "blk.31.attn_k_norm.weight": "bf16", |
| "blk.31.attn_output.weight": "i8*bf16[1x128]", |
| "blk.31.attn_q_norm.weight": "bf16", |
| "blk.31.attn_qg.weight": "i8*bf16[1x128]", |
| "blk.31.attn_v.weight": "i8*bf16[1x128]", |
| "blk.31.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.31.ffn_gate_inp.weight": "bf16", |
| "blk.31.ffn_gate_shexp.weight": "bf16", |
| "blk.31.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.31.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.31.ffn_hc_inject.weight": "bf16", |
| "blk.31.ffn_hc_norm.weight": "bf16", |
| "blk.31.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.31.qsa_k_norm.weight": "bf16", |
| "blk.31.qsa_q_norm.weight": "bf16", |
| "blk.31.qsa_qk.weight": "bf16", |
| "blk.32.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.32.attn_hc_inject.weight": "bf16", |
| "blk.32.attn_hc_norm.weight": "bf16", |
| "blk.32.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.32.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.32.ffn_gate_inp.weight": "bf16", |
| "blk.32.ffn_gate_shexp.weight": "bf16", |
| "blk.32.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.32.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.32.ffn_hc_inject.weight": "bf16", |
| "blk.32.ffn_hc_norm.weight": "bf16", |
| "blk.32.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.32.ssm_a_log": "bf16", |
| "blk.32.ssm_ab.weight": "bf16", |
| "blk.32.ssm_conv1d.weight": "bf16", |
| "blk.32.ssm_dt.bias": "bf16", |
| "blk.32.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.32.ssm_norm.weight": "bf16", |
| "blk.32.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.33.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.33.attn_hc_inject.weight": "bf16", |
| "blk.33.attn_hc_norm.weight": "bf16", |
| "blk.33.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.33.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.33.ffn_gate_inp.weight": "bf16", |
| "blk.33.ffn_gate_shexp.weight": "bf16", |
| "blk.33.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.33.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.33.ffn_hc_inject.weight": "bf16", |
| "blk.33.ffn_hc_norm.weight": "bf16", |
| "blk.33.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.33.ssm_a_log": "bf16", |
| "blk.33.ssm_ab.weight": "bf16", |
| "blk.33.ssm_conv1d.weight": "bf16", |
| "blk.33.ssm_dt.bias": "bf16", |
| "blk.33.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.33.ssm_norm.weight": "bf16", |
| "blk.33.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.34.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.34.attn_hc_inject.weight": "bf16", |
| "blk.34.attn_hc_norm.weight": "bf16", |
| "blk.34.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.34.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.34.ffn_gate_inp.weight": "bf16", |
| "blk.34.ffn_gate_shexp.weight": "bf16", |
| "blk.34.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.34.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.34.ffn_hc_inject.weight": "bf16", |
| "blk.34.ffn_hc_norm.weight": "bf16", |
| "blk.34.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.34.ssm_a_log": "bf16", |
| "blk.34.ssm_ab.weight": "bf16", |
| "blk.34.ssm_conv1d.weight": "bf16", |
| "blk.34.ssm_dt.bias": "bf16", |
| "blk.34.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.34.ssm_norm.weight": "bf16", |
| "blk.34.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.35.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.35.attn_hc_inject.weight": "bf16", |
| "blk.35.attn_hc_norm.weight": "bf16", |
| "blk.35.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.35.attn_k.weight": "i8*bf16[1x128]", |
| "blk.35.attn_k_norm.weight": "bf16", |
| "blk.35.attn_output.weight": "i8*bf16[1x128]", |
| "blk.35.attn_q_norm.weight": "bf16", |
| "blk.35.attn_qg.weight": "i8*bf16[1x128]", |
| "blk.35.attn_v.weight": "i8*bf16[1x128]", |
| "blk.35.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.35.ffn_gate_inp.weight": "bf16", |
| "blk.35.ffn_gate_shexp.weight": "bf16", |
| "blk.35.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.35.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.35.ffn_hc_inject.weight": "bf16", |
| "blk.35.ffn_hc_norm.weight": "bf16", |
| "blk.35.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.35.qsa_k_norm.weight": "bf16", |
| "blk.35.qsa_q_norm.weight": "bf16", |
| "blk.35.qsa_qk.weight": "bf16", |
| "blk.36.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.36.attn_hc_inject.weight": "bf16", |
| "blk.36.attn_hc_norm.weight": "bf16", |
| "blk.36.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.36.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.36.ffn_gate_inp.weight": "bf16", |
| "blk.36.ffn_gate_shexp.weight": "bf16", |
| "blk.36.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.36.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.36.ffn_hc_inject.weight": "bf16", |
| "blk.36.ffn_hc_norm.weight": "bf16", |
| "blk.36.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.36.ssm_a_log": "bf16", |
| "blk.36.ssm_ab.weight": "bf16", |
| "blk.36.ssm_conv1d.weight": "bf16", |
| "blk.36.ssm_dt.bias": "bf16", |
| "blk.36.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.36.ssm_norm.weight": "bf16", |
| "blk.36.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.37.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.37.attn_hc_inject.weight": "bf16", |
| "blk.37.attn_hc_norm.weight": "bf16", |
| "blk.37.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.37.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.37.ffn_gate_inp.weight": "bf16", |
| "blk.37.ffn_gate_shexp.weight": "bf16", |
| "blk.37.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.37.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.37.ffn_hc_inject.weight": "bf16", |
| "blk.37.ffn_hc_norm.weight": "bf16", |
| "blk.37.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.37.ssm_a_log": "bf16", |
| "blk.37.ssm_ab.weight": "bf16", |
| "blk.37.ssm_conv1d.weight": "bf16", |
| "blk.37.ssm_dt.bias": "bf16", |
| "blk.37.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.37.ssm_norm.weight": "bf16", |
| "blk.37.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.38.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.38.attn_hc_inject.weight": "bf16", |
| "blk.38.attn_hc_norm.weight": "bf16", |
| "blk.38.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.38.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.38.ffn_gate_inp.weight": "bf16", |
| "blk.38.ffn_gate_shexp.weight": "bf16", |
| "blk.38.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.38.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.38.ffn_hc_inject.weight": "bf16", |
| "blk.38.ffn_hc_norm.weight": "bf16", |
| "blk.38.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.38.ssm_a_log": "bf16", |
| "blk.38.ssm_ab.weight": "bf16", |
| "blk.38.ssm_conv1d.weight": "bf16", |
| "blk.38.ssm_dt.bias": "bf16", |
| "blk.38.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.38.ssm_norm.weight": "bf16", |
| "blk.38.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.39.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.39.attn_hc_inject.weight": "bf16", |
| "blk.39.attn_hc_norm.weight": "bf16", |
| "blk.39.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.39.attn_k.weight": "i8*bf16[1x128]", |
| "blk.39.attn_k_norm.weight": "bf16", |
| "blk.39.attn_output.weight": "i8*bf16[1x128]", |
| "blk.39.attn_q_norm.weight": "bf16", |
| "blk.39.attn_qg.weight": "i8*bf16[1x128]", |
| "blk.39.attn_v.weight": "i8*bf16[1x128]", |
| "blk.39.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.39.ffn_gate_inp.weight": "bf16", |
| "blk.39.ffn_gate_shexp.weight": "bf16", |
| "blk.39.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.39.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.39.ffn_hc_inject.weight": "bf16", |
| "blk.39.ffn_hc_norm.weight": "bf16", |
| "blk.39.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.39.qsa_k_norm.weight": "bf16", |
| "blk.39.qsa_q_norm.weight": "bf16", |
| "blk.39.qsa_qk.weight": "bf16", |
| "blk.40.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.40.attn_hc_inject.weight": "bf16", |
| "blk.40.attn_hc_norm.weight": "bf16", |
| "blk.40.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.40.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.40.ffn_gate_inp.weight": "bf16", |
| "blk.40.ffn_gate_shexp.weight": "bf16", |
| "blk.40.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.40.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.40.ffn_hc_inject.weight": "bf16", |
| "blk.40.ffn_hc_norm.weight": "bf16", |
| "blk.40.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.40.ssm_a_log": "bf16", |
| "blk.40.ssm_ab.weight": "bf16", |
| "blk.40.ssm_conv1d.weight": "bf16", |
| "blk.40.ssm_dt.bias": "bf16", |
| "blk.40.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.40.ssm_norm.weight": "bf16", |
| "blk.40.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.41.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.41.attn_hc_inject.weight": "bf16", |
| "blk.41.attn_hc_norm.weight": "bf16", |
| "blk.41.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.41.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.41.ffn_gate_inp.weight": "bf16", |
| "blk.41.ffn_gate_shexp.weight": "bf16", |
| "blk.41.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.41.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.41.ffn_hc_inject.weight": "bf16", |
| "blk.41.ffn_hc_norm.weight": "bf16", |
| "blk.41.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.41.ssm_a_log": "bf16", |
| "blk.41.ssm_ab.weight": "bf16", |
| "blk.41.ssm_conv1d.weight": "bf16", |
| "blk.41.ssm_dt.bias": "bf16", |
| "blk.41.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.41.ssm_norm.weight": "bf16", |
| "blk.41.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.42.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.42.attn_hc_inject.weight": "bf16", |
| "blk.42.attn_hc_norm.weight": "bf16", |
| "blk.42.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.42.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.42.ffn_gate_inp.weight": "bf16", |
| "blk.42.ffn_gate_shexp.weight": "bf16", |
| "blk.42.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.42.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.42.ffn_hc_inject.weight": "bf16", |
| "blk.42.ffn_hc_norm.weight": "bf16", |
| "blk.42.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.42.ssm_a_log": "bf16", |
| "blk.42.ssm_ab.weight": "bf16", |
| "blk.42.ssm_conv1d.weight": "bf16", |
| "blk.42.ssm_dt.bias": "bf16", |
| "blk.42.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.42.ssm_norm.weight": "bf16", |
| "blk.42.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.43.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.43.attn_hc_inject.weight": "bf16", |
| "blk.43.attn_hc_norm.weight": "bf16", |
| "blk.43.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.43.attn_k.weight": "i8*bf16[1x128]", |
| "blk.43.attn_k_norm.weight": "bf16", |
| "blk.43.attn_output.weight": "i8*bf16[1x128]", |
| "blk.43.attn_q_norm.weight": "bf16", |
| "blk.43.attn_qg.weight": "i8*bf16[1x128]", |
| "blk.43.attn_v.weight": "i8*bf16[1x128]", |
| "blk.43.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.43.ffn_gate_inp.weight": "bf16", |
| "blk.43.ffn_gate_shexp.weight": "bf16", |
| "blk.43.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.43.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.43.ffn_hc_inject.weight": "bf16", |
| "blk.43.ffn_hc_norm.weight": "bf16", |
| "blk.43.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.43.qsa_k_norm.weight": "bf16", |
| "blk.43.qsa_q_norm.weight": "bf16", |
| "blk.43.qsa_qk.weight": "bf16", |
| "blk.44.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.44.attn_hc_inject.weight": "bf16", |
| "blk.44.attn_hc_norm.weight": "bf16", |
| "blk.44.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.44.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.44.ffn_gate_inp.weight": "bf16", |
| "blk.44.ffn_gate_shexp.weight": "bf16", |
| "blk.44.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.44.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.44.ffn_hc_inject.weight": "bf16", |
| "blk.44.ffn_hc_norm.weight": "bf16", |
| "blk.44.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.44.ssm_a_log": "bf16", |
| "blk.44.ssm_ab.weight": "bf16", |
| "blk.44.ssm_conv1d.weight": "bf16", |
| "blk.44.ssm_dt.bias": "bf16", |
| "blk.44.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.44.ssm_norm.weight": "bf16", |
| "blk.44.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.45.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.45.attn_hc_inject.weight": "bf16", |
| "blk.45.attn_hc_norm.weight": "bf16", |
| "blk.45.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.45.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.45.ffn_gate_inp.weight": "bf16", |
| "blk.45.ffn_gate_shexp.weight": "bf16", |
| "blk.45.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.45.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.45.ffn_hc_inject.weight": "bf16", |
| "blk.45.ffn_hc_norm.weight": "bf16", |
| "blk.45.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.45.ssm_a_log": "bf16", |
| "blk.45.ssm_ab.weight": "bf16", |
| "blk.45.ssm_conv1d.weight": "bf16", |
| "blk.45.ssm_dt.bias": "bf16", |
| "blk.45.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.45.ssm_norm.weight": "bf16", |
| "blk.45.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.46.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.46.attn_hc_inject.weight": "bf16", |
| "blk.46.attn_hc_norm.weight": "bf16", |
| "blk.46.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.46.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.46.ffn_gate_inp.weight": "bf16", |
| "blk.46.ffn_gate_shexp.weight": "bf16", |
| "blk.46.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.46.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.46.ffn_hc_inject.weight": "bf16", |
| "blk.46.ffn_hc_norm.weight": "bf16", |
| "blk.46.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.46.ssm_a_log": "bf16", |
| "blk.46.ssm_ab.weight": "bf16", |
| "blk.46.ssm_conv1d.weight": "bf16", |
| "blk.46.ssm_dt.bias": "bf16", |
| "blk.46.ssm_inz.weight": "i8*bf16[1x128]", |
| "blk.46.ssm_norm.weight": "bf16", |
| "blk.46.ssm_out.weight": "i8*bf16[1x128]", |
| "blk.47.attn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.47.attn_hc_inject.weight": "bf16", |
| "blk.47.attn_hc_norm.weight": "bf16", |
| "blk.47.attn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.47.attn_k.weight": "i8*bf16[1x128]", |
| "blk.47.attn_k_norm.weight": "bf16", |
| "blk.47.attn_output.weight": "i8*bf16[1x128]", |
| "blk.47.attn_q_norm.weight": "bf16", |
| "blk.47.attn_qg.weight": "i8*bf16[1x128]", |
| "blk.47.attn_v.weight": "i8*bf16[1x128]", |
| "blk.47.ffn_down_shexp.weight": "i8*bf16[1x128]", |
| "blk.47.ffn_gate_inp.weight": "bf16", |
| "blk.47.ffn_gate_shexp.weight": "bf16", |
| "blk.47.ffn_gate_up_shexp.weight": "i8*bf16[1x128]", |
| "blk.47.ffn_hc_down.weight": "fp8_e4m3*f32[1x128]", |
| "blk.47.ffn_hc_inject.weight": "bf16", |
| "blk.47.ffn_hc_norm.weight": "bf16", |
| "blk.47.ffn_hc_up.weight": "fp8_e4m3*f32[1x80]", |
| "blk.47.qsa_k_norm.weight": "bf16", |
| "blk.47.qsa_q_norm.weight": "bf16", |
| "blk.47.qsa_qk.weight": "bf16" |
| }, |
| "anchors": { |
| "blk.23.attn_hc_norm.weight": "1c74a996abe9e943eec4ff8c1f83b042a0a372ef6477bdda5e0b95def861ccfd", |
| "blk.23.ffn_hc_norm.weight": "daf0e9488ec977d18bffe3c09cde15a4cde1d830411394dd59b1f96158e418cc", |
| "blk.24.attn_hc_norm.weight": "b2b1d65b2bfcea1e6a445981690283351699fcb944a3a54218f0f32d33f02b69" |
| } |
| }, |
| "split": 24, |
| "stream": "hc", |
| "stream_width": 10240, |
| "block_in_width": 2560, |
| "layers": { |
| "24": "recurrent", |
| "25": "recurrent", |
| "26": "recurrent", |
| "27": "full_attn", |
| "28": "recurrent", |
| "29": "recurrent", |
| "30": "recurrent", |
| "31": "full_attn", |
| "32": "recurrent", |
| "33": "recurrent", |
| "34": "recurrent", |
| "35": "full_attn", |
| "36": "recurrent", |
| "37": "recurrent", |
| "38": "recurrent", |
| "39": "full_attn", |
| "40": "recurrent", |
| "41": "recurrent", |
| "42": "recurrent", |
| "43": "full_attn", |
| "44": "recurrent", |
| "45": "recurrent", |
| "46": "recurrent", |
| "47": "full_attn" |
| }, |
| "tail": { |
| "default": 2048, |
| "min": 512, |
| "table": [ |
| 1024, |
| 2048, |
| 2560, |
| 3072 |
| ] |
| }, |
| "correction": { |
| "kind": "gdn_terminal_state", |
| "alpha_default": 1.0, |
| "alpha_table": [ |
| 0.0, |
| 0.5, |
| 1.0 |
| ], |
| "heads": 48, |
| "sd": [ |
| 128, |
| 128 |
| ], |
| "layers": [ |
| 24, |
| 25, |
| 26, |
| 28, |
| 29, |
| 30, |
| 32, |
| 33, |
| 34, |
| 36, |
| 37, |
| 38, |
| 40, |
| 41, |
| 42, |
| 44, |
| 45, |
| 46 |
| ] |
| }, |
| "rowsel": { |
| "classes": "cap,mixed,piece", |
| "share_default": 0.25, |
| "share_table": [ |
| 0.1, |
| 0.25, |
| 0.5 |
| ] |
| }, |
| "projector": { |
| "dtype": "i8", |
| "layout": "i8_row128", |
| "encoding": "i8*bf16[1x128]", |
| "files": { |
| "24": "proj8.L24.safetensors", |
| "25": "proj8.L25.safetensors", |
| "26": "proj8.L26.safetensors", |
| "27": "proj8.L27.safetensors", |
| "28": "proj8.L28.safetensors", |
| "29": "proj8.L29.safetensors", |
| "30": "proj8.L30.safetensors", |
| "31": "proj8.L31.safetensors", |
| "32": "proj8.L32.safetensors", |
| "33": "proj8.L33.safetensors", |
| "34": "proj8.L34.safetensors", |
| "35": "proj8.L35.safetensors", |
| "36": "proj8.L36.safetensors", |
| "37": "proj8.L37.safetensors", |
| "38": "proj8.L38.safetensors", |
| "39": "proj8.L39.safetensors", |
| "40": "proj8.L40.safetensors", |
| "41": "proj8.L41.safetensors", |
| "42": "proj8.L42.safetensors", |
| "43": "proj8.L43.safetensors", |
| "44": "proj8.L44.safetensors", |
| "45": "proj8.L45.safetensors", |
| "46": "proj8.L46.safetensors", |
| "47": "proj8.L47.safetensors" |
| }, |
| "source": { |
| "folder": "projector", |
| "manifest_sha256": "73ace68e581b833fc194cd9695369eac65b75c3a118ad467cc07871c3a10afe7" |
| } |
| }, |
| "marker": { |
| "spec": { |
| "comment": "Initial marker spec for Qwen3.8-Flash-Next (PLAN-FIX \u00a711.1/\u00a712). 64 tokens: offsets 0-59 fixed pattern, 60 ridgefill_share, 61 ridgefill_alpha, 62 ridgefill_tail, 63 end. The Stage F implementer may revise; the template tool must read everything from this file.", |
| "length": 64, |
| "pattern": { |
| "offsets": [ |
| 0, |
| 59 |
| ], |
| "tokens": [ |
| "<|quad_start|>", |
| "<|quad_end|>" |
| ], |
| "rule": "alternate, starting with the first token" |
| }, |
| "dials": [ |
| { |
| "offset": 60, |
| "kwarg": "ridgefill_share", |
| "default": "0.25", |
| "table": { |
| "0.10": "<|box_start|>", |
| "0.25": "<|box_end|>", |
| "0.50": "<|object_ref_start|>" |
| } |
| }, |
| { |
| "offset": 61, |
| "kwarg": "ridgefill_alpha", |
| "default": "1.0", |
| "table": { |
| "0": "<|box_start|>", |
| "0.5": "<|box_end|>", |
| "1.0": "<|object_ref_start|>" |
| } |
| }, |
| { |
| "offset": 62, |
| "kwarg": "ridgefill_tail", |
| "default": "2048", |
| "table": { |
| "1024": "<|box_start|>", |
| "2048": "<|box_end|>", |
| "2560": "<|object_ref_start|>", |
| "3072": "<|object_ref_end|>" |
| } |
| } |
| ], |
| "end": { |
| "offset": 63, |
| "token": "<|endoftext|>" |
| }, |
| "switch": { |
| "kwarg": "ridgefill", |
| "on_value": "on" |
| }, |
| "token_ids": { |
| "<|quad_start|>": 248051, |
| "<|quad_end|>": 248052, |
| "<|box_start|>": 248049, |
| "<|box_end|>": 248050, |
| "<|object_ref_start|>": 248047, |
| "<|object_ref_end|>": 248048, |
| "<|endoftext|>": 248044 |
| } |
| }, |
| "template_source_sha256": "c3cf9e34abf4f9e36c2d72165aa9c132d3e2a725b6c2586aaa3a8af9d7a81041" |
| }, |
| "fit": { |
| "engine": "radiance 1.0.8", |
| "proj_sha256": "c9db6064eb275d3166df8139239959dfdeb64f2778707a2f42215164999cb9cd", |
| "st0_sha256": "38e4b550dfc1419922feccb59f75f758be7e8f2468c84b51a30712914b0b5dd3", |
| "st1_sha256": "af517cf760f7baef3f514c48ec4ab36d31effa444641cd6a0f87fdae2e384a18", |
| "freq_sha256": "4f1b719a59fdaa6ddfc7e06502b817c5a96e6a00bca060136d110e672059f97a", |
| "tokenizer_json_sha256": "0997f410c57a1f4e53b09e4be8f4a172d90edd9564368fb0847030937229b9f3" |
| }, |
| "files": { |
| "chat_template.jinja": "3c1c94830884ec96c61733698f7af4af088900e8852a2c2c385a779bb5d4c9fe", |
| "correction.safetensors": "d63ecba03570cccd84a2f0e966c98f5c2f2a2053bddcb37c85449808830f52bc", |
| "proj8.L24.safetensors": "523f6aa12b839fefac1ed532d8810fa284ab0bfe6da988e8b8e2bd87cbbaac96", |
| "proj8.L25.safetensors": "017926e427fa0328dcb8116b7f390de5015654f413055dfb0f0a01071a5c2b56", |
| "proj8.L26.safetensors": "598b70a678336d5222c0b1b15b966f1aec934160bbd0c438047381f2a5f89936", |
| "proj8.L27.safetensors": "45948fd9d4c2f15822703c0c9c28a2a3ba545700dbf33827379f4b60824dfc68", |
| "proj8.L28.safetensors": "40c547c673594ec525f0075a19ca4e455df7815ca4f373e8ac7078c49d785a8a", |
| "proj8.L29.safetensors": "0b93b1116d9edadc7fbe4cd1c13d7034e39ceb4f0a936961025807c843c5afe2", |
| "proj8.L30.safetensors": "fde7f4b1c3d3b9f1ded601e6708285af3b37e0c7a9baf3f6b4a137f05ba7b0ea", |
| "proj8.L31.safetensors": "3efc9863d8eb459853295c93c964c7634b14a006788142bcf2b94de3decc3587", |
| "proj8.L32.safetensors": "79eba94ca8336c83d2a831b5181172bf3da137ea7ca73719afa33342904dd201", |
| "proj8.L33.safetensors": "8973a39b03ab7ad7cd975ad2a69799ed2fecbdda7095d3194fe82f29e98731ff", |
| "proj8.L34.safetensors": "ef2b45cc049826cfb86af998df5bc0221273246b557b55f5259f5ee5664305c8", |
| "proj8.L35.safetensors": "9b26d52d1210f80611e69c46e1865f28251a0b629c2eac81a4b9475a9604e00e", |
| "proj8.L36.safetensors": "a869ae2d7e97e9f295cb21345fbff5c7a90c1a2a3e8d81ab46a98e12134a492f", |
| "proj8.L37.safetensors": "6e11339f558c5769b8c97d0dba5745fc5320871f09d0cefc7326be83b729b087", |
| "proj8.L38.safetensors": "69d1839b1ae123aeeca927250843d0dc86cde3b5dab28540ea8306ff62222708", |
| "proj8.L39.safetensors": "7d0394373111bf678f93a6cadcb0c00c9f1eb5d0013d3e70289828318728e15d", |
| "proj8.L40.safetensors": "345bbb651bf9764db82fa958716504fbd203ca612a6663231c91b9ee28bbc3b6", |
| "proj8.L41.safetensors": "eb4790257e06c9d1588ff84743d644dc83097bc71bb47e37f5e119fac42579d3", |
| "proj8.L42.safetensors": "6007d7dbf9904a00a55684660b3a4fc2af85ac65b60f796aae3c79b8878c161a", |
| "proj8.L43.safetensors": "3e0a1e52f91ffe049cd17aa269e2a16230e9b1baa88edfec1eecb437b192f187", |
| "proj8.L44.safetensors": "a2975e6edeb5165fc4ed4d095dd20f519a7f409163b08fbe3799202bbef29da3", |
| "proj8.L45.safetensors": "262666eecc1a40ce07a6af538390a523c9230095444fc640f0199f049a5339a0", |
| "proj8.L46.safetensors": "37b97f234a870320d1b3646086caa68358c4483b37c60cab940a23db57f6ad6f", |
| "proj8.L47.safetensors": "e04d881074f100bc271c61d93867677110041bfca8e7f0257835690cb515306a", |
| "rowsel.safetensors": "272414e5e19912ba8a93199558dbe7874ace73a382e7ed500ed16f6d566e29f3" |
| }, |
| "rebuilt_from": { |
| "folder": "projector-qwen38fn-int8", |
| "manifest_sha256": "8b4fc11340aeeec8560f7973786b8f6c44f555f920f10c06299cb2512a2f6308" |
| } |
| } |
|
|
