diff --git a/config.json b/config.json
index 05b4ca4a619711d812cd5b11d2581dd5e53fbca6..0da270ab240cd15c1a112d2095407dd544777972 100644
--- a/config.json
+++ b/config.json
@@ -1,3674 +1,2987 @@
 {
-  "_name_or_path": "",
   "architectures": [
     "Glm5NextForConditionalGeneration"
   ],
-  "chunk_size_feed_forward": 0,
-  "dtype": "bfloat16",
-  "id2label": {
-    "0": "LABEL_0",
-    "1": "LABEL_1"
-  },
   "image_end_token_id": 154831,
   "image_start_token_id": 154830,
   "image_token_id": 154854,
-  "is_encoder_decoder": false,
-  "label2id": {
-    "LABEL_0": 0,
-    "LABEL_1": 1
-  },
   "model_type": "glm5_next",
-  "output_attentions": false,
-  "output_hidden_states": false,
-  "problem_type": null,
   "quantization_config": {
-    "config_groups": {
-      "group_nvfp4_routed_experts": {
-        "input_activations": {
-          "dynamic": false,
-          "group_size": 16,
-          "num_bits": 4,
-          "type": "float"
-        },
-        "targets": [
-          "model.language_model.layers.3.mlp.experts",
-          "model.language_model.layers.4.mlp.experts",
-          "model.language_model.layers.5.mlp.experts",
-          "model.language_model.layers.6.mlp.experts",
-          "model.language_model.layers.7.mlp.experts",
-          "model.language_model.layers.8.mlp.experts",
-          "model.language_model.layers.9.mlp.experts",
-          "model.language_model.layers.10.mlp.experts",
-          "model.language_model.layers.11.mlp.experts",
-          "model.language_model.layers.12.mlp.experts",
-          "model.language_model.layers.13.mlp.experts",
-          "model.language_model.layers.14.mlp.experts",
-          "model.language_model.layers.15.mlp.experts",
-          "model.language_model.layers.16.mlp.experts",
-          "model.language_model.layers.17.mlp.experts",
-          "model.language_model.layers.18.mlp.experts",
-          "model.language_model.layers.19.mlp.experts",
-          "model.language_model.layers.20.mlp.experts",
-          "model.language_model.layers.21.mlp.experts",
-          "model.language_model.layers.22.mlp.experts",
-          "model.language_model.layers.23.mlp.experts",
-          "model.language_model.layers.24.mlp.experts",
-          "model.language_model.layers.25.mlp.experts",
-          "model.language_model.layers.26.mlp.experts",
-          "model.language_model.layers.27.mlp.experts",
-          "model.language_model.layers.28.mlp.experts",
-          "model.language_model.layers.29.mlp.experts",
-          "model.language_model.layers.30.mlp.experts",
-          "model.language_model.layers.31.mlp.experts",
-          "model.language_model.layers.32.mlp.experts",
-          "model.language_model.layers.33.mlp.experts",
-          "model.language_model.layers.34.mlp.experts",
-          "model.language_model.layers.35.mlp.experts",
-          "model.language_model.layers.36.mlp.experts",
-          "model.language_model.layers.37.mlp.experts",
-          "model.language_model.layers.38.mlp.experts",
-          "model.language_model.layers.39.mlp.experts",
-          "model.language_model.layers.40.mlp.experts",
-          "model.language_model.layers.41.mlp.experts",
-          "model.language_model.layers.42.mlp.experts",
-          "model.language_model.layers.43.mlp.experts",
-          "model.language_model.layers.44.mlp.experts"
-        ],
-        "weights": {
-          "dynamic": false,
-          "group_size": 16,
-          "num_bits": 4,
-          "type": "float"
-        }
-      },
-      "group_mxfp8_attention": {
-        "input_activations": {
-          "dynamic": true,
-          "num_bits": 8,
-          "type": "float",
-          "group_size": 32
-        },
-        "weights": {
-          "dynamic": false,
-          "num_bits": 8,
-          "type": "float",
-          "group_size": 32,
-          "weight_block_size": [
-            1,
-            32
-          ]
-        },
-        "targets": [
-          "model.language_model.layers.0.self_attn.b_proj",
-          "model.language_model.layers.0.self_attn.f_a_proj",
-          "model.language_model.layers.0.self_attn.f_b_proj",
-          "model.language_model.layers.0.self_attn.g_a_proj",
-          "model.language_model.layers.0.self_attn.g_b_proj",
-          "model.language_model.layers.0.self_attn.k_proj",
-          "model.language_model.layers.0.self_attn.o_proj",
-          "model.language_model.layers.0.self_attn.q_proj",
-          "model.language_model.layers.0.self_attn.v_proj",
-          "model.language_model.layers.1.self_attn.b_proj",
-          "model.language_model.layers.1.self_attn.f_a_proj",
-          "model.language_model.layers.1.self_attn.f_b_proj",
-          "model.language_model.layers.1.self_attn.g_a_proj",
-          "model.language_model.layers.1.self_attn.g_b_proj",
-          "model.language_model.layers.1.self_attn.k_proj",
-          "model.language_model.layers.1.self_attn.o_proj",
-          "model.language_model.layers.1.self_attn.q_proj",
-          "model.language_model.layers.1.self_attn.v_proj",
-          "model.language_model.layers.10.self_attn.b_proj",
-          "model.language_model.layers.10.self_attn.f_a_proj",
-          "model.language_model.layers.10.self_attn.f_b_proj",
-          "model.language_model.layers.10.self_attn.g_a_proj",
-          "model.language_model.layers.10.self_attn.g_b_proj",
-          "model.language_model.layers.10.self_attn.k_proj",
-          "model.language_model.layers.10.self_attn.o_proj",
-          "model.language_model.layers.10.self_attn.q_proj",
-          "model.language_model.layers.10.self_attn.v_proj",
-          "model.language_model.layers.11.self_attn.indexer.weights_proj",
-          "model.language_model.layers.11.self_attn.indexer.wk",
-          "model.language_model.layers.11.self_attn.indexer.wq_b",
-          "model.language_model.layers.11.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.11.self_attn.kv_b_proj",
-          "model.language_model.layers.11.self_attn.o_proj",
-          "model.language_model.layers.11.self_attn.q_a_proj",
-          "model.language_model.layers.11.self_attn.q_b_proj",
-          "model.language_model.layers.12.self_attn.b_proj",
-          "model.language_model.layers.12.self_attn.f_a_proj",
-          "model.language_model.layers.12.self_attn.f_b_proj",
-          "model.language_model.layers.12.self_attn.g_a_proj",
-          "model.language_model.layers.12.self_attn.g_b_proj",
-          "model.language_model.layers.12.self_attn.k_proj",
-          "model.language_model.layers.12.self_attn.o_proj",
-          "model.language_model.layers.12.self_attn.q_proj",
-          "model.language_model.layers.12.self_attn.v_proj",
-          "model.language_model.layers.13.self_attn.b_proj",
-          "model.language_model.layers.13.self_attn.f_a_proj",
-          "model.language_model.layers.13.self_attn.f_b_proj",
-          "model.language_model.layers.13.self_attn.g_a_proj",
-          "model.language_model.layers.13.self_attn.g_b_proj",
-          "model.language_model.layers.13.self_attn.k_proj",
-          "model.language_model.layers.13.self_attn.o_proj",
-          "model.language_model.layers.13.self_attn.q_proj",
-          "model.language_model.layers.13.self_attn.v_proj",
-          "model.language_model.layers.14.self_attn.b_proj",
-          "model.language_model.layers.14.self_attn.f_a_proj",
-          "model.language_model.layers.14.self_attn.f_b_proj",
-          "model.language_model.layers.14.self_attn.g_a_proj",
-          "model.language_model.layers.14.self_attn.g_b_proj",
-          "model.language_model.layers.14.self_attn.k_proj",
-          "model.language_model.layers.14.self_attn.o_proj",
-          "model.language_model.layers.14.self_attn.q_proj",
-          "model.language_model.layers.14.self_attn.v_proj",
-          "model.language_model.layers.15.self_attn.indexer.weights_proj",
-          "model.language_model.layers.15.self_attn.indexer.wk",
-          "model.language_model.layers.15.self_attn.indexer.wq_b",
-          "model.language_model.layers.15.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.15.self_attn.kv_b_proj",
-          "model.language_model.layers.15.self_attn.o_proj",
-          "model.language_model.layers.15.self_attn.q_a_proj",
-          "model.language_model.layers.15.self_attn.q_b_proj",
-          "model.language_model.layers.16.self_attn.b_proj",
-          "model.language_model.layers.16.self_attn.f_a_proj",
-          "model.language_model.layers.16.self_attn.f_b_proj",
-          "model.language_model.layers.16.self_attn.g_a_proj",
-          "model.language_model.layers.16.self_attn.g_b_proj",
-          "model.language_model.layers.16.self_attn.k_proj",
-          "model.language_model.layers.16.self_attn.o_proj",
-          "model.language_model.layers.16.self_attn.q_proj",
-          "model.language_model.layers.16.self_attn.v_proj",
-          "model.language_model.layers.17.self_attn.b_proj",
-          "model.language_model.layers.17.self_attn.f_a_proj",
-          "model.language_model.layers.17.self_attn.f_b_proj",
-          "model.language_model.layers.17.self_attn.g_a_proj",
-          "model.language_model.layers.17.self_attn.g_b_proj",
-          "model.language_model.layers.17.self_attn.k_proj",
-          "model.language_model.layers.17.self_attn.o_proj",
-          "model.language_model.layers.17.self_attn.q_proj",
-          "model.language_model.layers.17.self_attn.v_proj",
-          "model.language_model.layers.18.self_attn.b_proj",
-          "model.language_model.layers.18.self_attn.f_a_proj",
-          "model.language_model.layers.18.self_attn.f_b_proj",
-          "model.language_model.layers.18.self_attn.g_a_proj",
-          "model.language_model.layers.18.self_attn.g_b_proj",
-          "model.language_model.layers.18.self_attn.k_proj",
-          "model.language_model.layers.18.self_attn.o_proj",
-          "model.language_model.layers.18.self_attn.q_proj",
-          "model.language_model.layers.18.self_attn.v_proj",
-          "model.language_model.layers.19.self_attn.indexer.weights_proj",
-          "model.language_model.layers.19.self_attn.indexer.wk",
-          "model.language_model.layers.19.self_attn.indexer.wq_b",
-          "model.language_model.layers.19.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.19.self_attn.kv_b_proj",
-          "model.language_model.layers.19.self_attn.o_proj",
-          "model.language_model.layers.19.self_attn.q_a_proj",
-          "model.language_model.layers.19.self_attn.q_b_proj",
-          "model.language_model.layers.2.self_attn.b_proj",
-          "model.language_model.layers.2.self_attn.f_a_proj",
-          "model.language_model.layers.2.self_attn.f_b_proj",
-          "model.language_model.layers.2.self_attn.g_a_proj",
-          "model.language_model.layers.2.self_attn.g_b_proj",
-          "model.language_model.layers.2.self_attn.k_proj",
-          "model.language_model.layers.2.self_attn.o_proj",
-          "model.language_model.layers.2.self_attn.q_proj",
-          "model.language_model.layers.2.self_attn.v_proj",
-          "model.language_model.layers.20.self_attn.b_proj",
-          "model.language_model.layers.20.self_attn.f_a_proj",
-          "model.language_model.layers.20.self_attn.f_b_proj",
-          "model.language_model.layers.20.self_attn.g_a_proj",
-          "model.language_model.layers.20.self_attn.g_b_proj",
-          "model.language_model.layers.20.self_attn.k_proj",
-          "model.language_model.layers.20.self_attn.o_proj",
-          "model.language_model.layers.20.self_attn.q_proj",
-          "model.language_model.layers.20.self_attn.v_proj",
-          "model.language_model.layers.21.self_attn.b_proj",
-          "model.language_model.layers.21.self_attn.f_a_proj",
-          "model.language_model.layers.21.self_attn.f_b_proj",
-          "model.language_model.layers.21.self_attn.g_a_proj",
-          "model.language_model.layers.21.self_attn.g_b_proj",
-          "model.language_model.layers.21.self_attn.k_proj",
-          "model.language_model.layers.21.self_attn.o_proj",
-          "model.language_model.layers.21.self_attn.q_proj",
-          "model.language_model.layers.21.self_attn.v_proj",
-          "model.language_model.layers.22.self_attn.b_proj",
-          "model.language_model.layers.22.self_attn.f_a_proj",
-          "model.language_model.layers.22.self_attn.f_b_proj",
-          "model.language_model.layers.22.self_attn.g_a_proj",
-          "model.language_model.layers.22.self_attn.g_b_proj",
-          "model.language_model.layers.22.self_attn.k_proj",
-          "model.language_model.layers.22.self_attn.o_proj",
-          "model.language_model.layers.22.self_attn.q_proj",
-          "model.language_model.layers.22.self_attn.v_proj",
-          "model.language_model.layers.23.self_attn.indexer.weights_proj",
-          "model.language_model.layers.23.self_attn.indexer.wk",
-          "model.language_model.layers.23.self_attn.indexer.wq_b",
-          "model.language_model.layers.23.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.23.self_attn.kv_b_proj",
-          "model.language_model.layers.23.self_attn.o_proj",
-          "model.language_model.layers.23.self_attn.q_a_proj",
-          "model.language_model.layers.23.self_attn.q_b_proj",
-          "model.language_model.layers.24.self_attn.b_proj",
-          "model.language_model.layers.24.self_attn.f_a_proj",
-          "model.language_model.layers.24.self_attn.f_b_proj",
-          "model.language_model.layers.24.self_attn.g_a_proj",
-          "model.language_model.layers.24.self_attn.g_b_proj",
-          "model.language_model.layers.24.self_attn.k_proj",
-          "model.language_model.layers.24.self_attn.o_proj",
-          "model.language_model.layers.24.self_attn.q_proj",
-          "model.language_model.layers.24.self_attn.v_proj",
-          "model.language_model.layers.25.self_attn.b_proj",
-          "model.language_model.layers.25.self_attn.f_a_proj",
-          "model.language_model.layers.25.self_attn.f_b_proj",
-          "model.language_model.layers.25.self_attn.g_a_proj",
-          "model.language_model.layers.25.self_attn.g_b_proj",
-          "model.language_model.layers.25.self_attn.k_proj",
-          "model.language_model.layers.25.self_attn.o_proj",
-          "model.language_model.layers.25.self_attn.q_proj",
-          "model.language_model.layers.25.self_attn.v_proj",
-          "model.language_model.layers.26.self_attn.b_proj",
-          "model.language_model.layers.26.self_attn.f_a_proj",
-          "model.language_model.layers.26.self_attn.f_b_proj",
-          "model.language_model.layers.26.self_attn.g_a_proj",
-          "model.language_model.layers.26.self_attn.g_b_proj",
-          "model.language_model.layers.26.self_attn.k_proj",
-          "model.language_model.layers.26.self_attn.o_proj",
-          "model.language_model.layers.26.self_attn.q_proj",
-          "model.language_model.layers.26.self_attn.v_proj",
-          "model.language_model.layers.27.self_attn.indexer.weights_proj",
-          "model.language_model.layers.27.self_attn.indexer.wk",
-          "model.language_model.layers.27.self_attn.indexer.wq_b",
-          "model.language_model.layers.27.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.27.self_attn.kv_b_proj",
-          "model.language_model.layers.27.self_attn.o_proj",
-          "model.language_model.layers.27.self_attn.q_a_proj",
-          "model.language_model.layers.27.self_attn.q_b_proj",
-          "model.language_model.layers.28.self_attn.b_proj",
-          "model.language_model.layers.28.self_attn.f_a_proj",
-          "model.language_model.layers.28.self_attn.f_b_proj",
-          "model.language_model.layers.28.self_attn.g_a_proj",
-          "model.language_model.layers.28.self_attn.g_b_proj",
-          "model.language_model.layers.28.self_attn.k_proj",
-          "model.language_model.layers.28.self_attn.o_proj",
-          "model.language_model.layers.28.self_attn.q_proj",
-          "model.language_model.layers.28.self_attn.v_proj",
-          "model.language_model.layers.29.self_attn.b_proj",
-          "model.language_model.layers.29.self_attn.f_a_proj",
-          "model.language_model.layers.29.self_attn.f_b_proj",
-          "model.language_model.layers.29.self_attn.g_a_proj",
-          "model.language_model.layers.29.self_attn.g_b_proj",
-          "model.language_model.layers.29.self_attn.k_proj",
-          "model.language_model.layers.29.self_attn.o_proj",
-          "model.language_model.layers.29.self_attn.q_proj",
-          "model.language_model.layers.29.self_attn.v_proj",
-          "model.language_model.layers.3.self_attn.indexer.weights_proj",
-          "model.language_model.layers.3.self_attn.indexer.wk",
-          "model.language_model.layers.3.self_attn.indexer.wq_b",
-          "model.language_model.layers.3.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.3.self_attn.kv_b_proj",
-          "model.language_model.layers.3.self_attn.o_proj",
-          "model.language_model.layers.3.self_attn.q_a_proj",
-          "model.language_model.layers.3.self_attn.q_b_proj",
-          "model.language_model.layers.30.self_attn.b_proj",
-          "model.language_model.layers.30.self_attn.f_a_proj",
-          "model.language_model.layers.30.self_attn.f_b_proj",
-          "model.language_model.layers.30.self_attn.g_a_proj",
-          "model.language_model.layers.30.self_attn.g_b_proj",
-          "model.language_model.layers.30.self_attn.k_proj",
-          "model.language_model.layers.30.self_attn.o_proj",
-          "model.language_model.layers.30.self_attn.q_proj",
-          "model.language_model.layers.30.self_attn.v_proj",
-          "model.language_model.layers.31.self_attn.indexer.weights_proj",
-          "model.language_model.layers.31.self_attn.indexer.wk",
-          "model.language_model.layers.31.self_attn.indexer.wq_b",
-          "model.language_model.layers.31.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.31.self_attn.kv_b_proj",
-          "model.language_model.layers.31.self_attn.o_proj",
-          "model.language_model.layers.31.self_attn.q_a_proj",
-          "model.language_model.layers.31.self_attn.q_b_proj",
-          "model.language_model.layers.32.self_attn.b_proj",
-          "model.language_model.layers.32.self_attn.f_a_proj",
-          "model.language_model.layers.32.self_attn.f_b_proj",
-          "model.language_model.layers.32.self_attn.g_a_proj",
-          "model.language_model.layers.32.self_attn.g_b_proj",
-          "model.language_model.layers.32.self_attn.k_proj",
-          "model.language_model.layers.32.self_attn.o_proj",
-          "model.language_model.layers.32.self_attn.q_proj",
-          "model.language_model.layers.32.self_attn.v_proj",
-          "model.language_model.layers.33.self_attn.b_proj",
-          "model.language_model.layers.33.self_attn.f_a_proj",
-          "model.language_model.layers.33.self_attn.f_b_proj",
-          "model.language_model.layers.33.self_attn.g_a_proj",
-          "model.language_model.layers.33.self_attn.g_b_proj",
-          "model.language_model.layers.33.self_attn.k_proj",
-          "model.language_model.layers.33.self_attn.o_proj",
-          "model.language_model.layers.33.self_attn.q_proj",
-          "model.language_model.layers.33.self_attn.v_proj",
-          "model.language_model.layers.34.self_attn.b_proj",
-          "model.language_model.layers.34.self_attn.f_a_proj",
-          "model.language_model.layers.34.self_attn.f_b_proj",
-          "model.language_model.layers.34.self_attn.g_a_proj",
-          "model.language_model.layers.34.self_attn.g_b_proj",
-          "model.language_model.layers.34.self_attn.k_proj",
-          "model.language_model.layers.34.self_attn.o_proj",
-          "model.language_model.layers.34.self_attn.q_proj",
-          "model.language_model.layers.34.self_attn.v_proj",
-          "model.language_model.layers.35.self_attn.indexer.weights_proj",
-          "model.language_model.layers.35.self_attn.indexer.wk",
-          "model.language_model.layers.35.self_attn.indexer.wq_b",
-          "model.language_model.layers.35.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.35.self_attn.kv_b_proj",
-          "model.language_model.layers.35.self_attn.o_proj",
-          "model.language_model.layers.35.self_attn.q_a_proj",
-          "model.language_model.layers.35.self_attn.q_b_proj",
-          "model.language_model.layers.36.self_attn.b_proj",
-          "model.language_model.layers.36.self_attn.f_a_proj",
-          "model.language_model.layers.36.self_attn.f_b_proj",
-          "model.language_model.layers.36.self_attn.g_a_proj",
-          "model.language_model.layers.36.self_attn.g_b_proj",
-          "model.language_model.layers.36.self_attn.k_proj",
-          "model.language_model.layers.36.self_attn.o_proj",
-          "model.language_model.layers.36.self_attn.q_proj",
-          "model.language_model.layers.36.self_attn.v_proj",
-          "model.language_model.layers.37.self_attn.b_proj",
-          "model.language_model.layers.37.self_attn.f_a_proj",
-          "model.language_model.layers.37.self_attn.f_b_proj",
-          "model.language_model.layers.37.self_attn.g_a_proj",
-          "model.language_model.layers.37.self_attn.g_b_proj",
-          "model.language_model.layers.37.self_attn.k_proj",
-          "model.language_model.layers.37.self_attn.o_proj",
-          "model.language_model.layers.37.self_attn.q_proj",
-          "model.language_model.layers.37.self_attn.v_proj",
-          "model.language_model.layers.38.self_attn.b_proj",
-          "model.language_model.layers.38.self_attn.f_a_proj",
-          "model.language_model.layers.38.self_attn.f_b_proj",
-          "model.language_model.layers.38.self_attn.g_a_proj",
-          "model.language_model.layers.38.self_attn.g_b_proj",
-          "model.language_model.layers.38.self_attn.k_proj",
-          "model.language_model.layers.38.self_attn.o_proj",
-          "model.language_model.layers.38.self_attn.q_proj",
-          "model.language_model.layers.38.self_attn.v_proj",
-          "model.language_model.layers.39.self_attn.indexer.weights_proj",
-          "model.language_model.layers.39.self_attn.indexer.wk",
-          "model.language_model.layers.39.self_attn.indexer.wq_b",
-          "model.language_model.layers.39.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.39.self_attn.kv_b_proj",
-          "model.language_model.layers.39.self_attn.o_proj",
-          "model.language_model.layers.39.self_attn.q_a_proj",
-          "model.language_model.layers.39.self_attn.q_b_proj",
-          "model.language_model.layers.4.self_attn.b_proj",
-          "model.language_model.layers.4.self_attn.f_a_proj",
-          "model.language_model.layers.4.self_attn.f_b_proj",
-          "model.language_model.layers.4.self_attn.g_a_proj",
-          "model.language_model.layers.4.self_attn.g_b_proj",
-          "model.language_model.layers.4.self_attn.k_proj",
-          "model.language_model.layers.4.self_attn.o_proj",
-          "model.language_model.layers.4.self_attn.q_proj",
-          "model.language_model.layers.4.self_attn.v_proj",
-          "model.language_model.layers.40.self_attn.b_proj",
-          "model.language_model.layers.40.self_attn.f_a_proj",
-          "model.language_model.layers.40.self_attn.f_b_proj",
-          "model.language_model.layers.40.self_attn.g_a_proj",
-          "model.language_model.layers.40.self_attn.g_b_proj",
-          "model.language_model.layers.40.self_attn.k_proj",
-          "model.language_model.layers.40.self_attn.o_proj",
-          "model.language_model.layers.40.self_attn.q_proj",
-          "model.language_model.layers.40.self_attn.v_proj",
-          "model.language_model.layers.41.self_attn.b_proj",
-          "model.language_model.layers.41.self_attn.f_a_proj",
-          "model.language_model.layers.41.self_attn.f_b_proj",
-          "model.language_model.layers.41.self_attn.g_a_proj",
-          "model.language_model.layers.41.self_attn.g_b_proj",
-          "model.language_model.layers.41.self_attn.k_proj",
-          "model.language_model.layers.41.self_attn.o_proj",
-          "model.language_model.layers.41.self_attn.q_proj",
-          "model.language_model.layers.41.self_attn.v_proj",
-          "model.language_model.layers.42.self_attn.b_proj",
-          "model.language_model.layers.42.self_attn.f_a_proj",
-          "model.language_model.layers.42.self_attn.f_b_proj",
-          "model.language_model.layers.42.self_attn.g_a_proj",
-          "model.language_model.layers.42.self_attn.g_b_proj",
-          "model.language_model.layers.42.self_attn.k_proj",
-          "model.language_model.layers.42.self_attn.o_proj",
-          "model.language_model.layers.42.self_attn.q_proj",
-          "model.language_model.layers.42.self_attn.v_proj",
-          "model.language_model.layers.43.self_attn.indexer.weights_proj",
-          "model.language_model.layers.43.self_attn.indexer.wk",
-          "model.language_model.layers.43.self_attn.indexer.wq_b",
-          "model.language_model.layers.43.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.43.self_attn.kv_b_proj",
-          "model.language_model.layers.43.self_attn.o_proj",
-          "model.language_model.layers.43.self_attn.q_a_proj",
-          "model.language_model.layers.43.self_attn.q_b_proj",
-          "model.language_model.layers.44.self_attn.b_proj",
-          "model.language_model.layers.44.self_attn.f_a_proj",
-          "model.language_model.layers.44.self_attn.f_b_proj",
-          "model.language_model.layers.44.self_attn.g_a_proj",
-          "model.language_model.layers.44.self_attn.g_b_proj",
-          "model.language_model.layers.44.self_attn.k_proj",
-          "model.language_model.layers.44.self_attn.o_proj",
-          "model.language_model.layers.44.self_attn.q_proj",
-          "model.language_model.layers.44.self_attn.v_proj",
-          "model.language_model.layers.45.self_attn.indexer.weights_proj",
-          "model.language_model.layers.45.self_attn.indexer.wk",
-          "model.language_model.layers.45.self_attn.indexer.wq_b",
-          "model.language_model.layers.45.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.45.self_attn.kv_b_proj",
-          "model.language_model.layers.45.self_attn.o_proj",
-          "model.language_model.layers.45.self_attn.q_a_proj",
-          "model.language_model.layers.45.self_attn.q_b_proj",
-          "model.language_model.layers.5.self_attn.b_proj",
-          "model.language_model.layers.5.self_attn.f_a_proj",
-          "model.language_model.layers.5.self_attn.f_b_proj",
-          "model.language_model.layers.5.self_attn.g_a_proj",
-          "model.language_model.layers.5.self_attn.g_b_proj",
-          "model.language_model.layers.5.self_attn.k_proj",
-          "model.language_model.layers.5.self_attn.o_proj",
-          "model.language_model.layers.5.self_attn.q_proj",
-          "model.language_model.layers.5.self_attn.v_proj",
-          "model.language_model.layers.6.self_attn.b_proj",
-          "model.language_model.layers.6.self_attn.f_a_proj",
-          "model.language_model.layers.6.self_attn.f_b_proj",
-          "model.language_model.layers.6.self_attn.g_a_proj",
-          "model.language_model.layers.6.self_attn.g_b_proj",
-          "model.language_model.layers.6.self_attn.k_proj",
-          "model.language_model.layers.6.self_attn.o_proj",
-          "model.language_model.layers.6.self_attn.q_proj",
-          "model.language_model.layers.6.self_attn.v_proj",
-          "model.language_model.layers.7.self_attn.indexer.weights_proj",
-          "model.language_model.layers.7.self_attn.indexer.wk",
-          "model.language_model.layers.7.self_attn.indexer.wq_b",
-          "model.language_model.layers.7.self_attn.kv_a_proj_with_mqa",
-          "model.language_model.layers.7.self_attn.kv_b_proj",
-          "model.language_model.layers.7.self_attn.o_proj",
-          "model.language_model.layers.7.self_attn.q_a_proj",
-          "model.language_model.layers.7.self_attn.q_b_proj",
-          "model.language_model.layers.8.self_attn.b_proj",
-          "model.language_model.layers.8.self_attn.f_a_proj",
-          "model.language_model.layers.8.self_attn.f_b_proj",
-          "model.language_model.layers.8.self_attn.g_a_proj",
-          "model.language_model.layers.8.self_attn.g_b_proj",
-          "model.language_model.layers.8.self_attn.k_proj",
-          "model.language_model.layers.8.self_attn.o_proj",
-          "model.language_model.layers.8.self_attn.q_proj",
-          "model.language_model.layers.8.self_attn.v_proj",
-          "model.language_model.layers.9.self_attn.b_proj",
-          "model.language_model.layers.9.self_attn.f_a_proj",
-          "model.language_model.layers.9.self_attn.f_b_proj",
-          "model.language_model.layers.9.self_attn.g_a_proj",
-          "model.language_model.layers.9.self_attn.g_b_proj",
-          "model.language_model.layers.9.self_attn.k_proj",
-          "model.language_model.layers.9.self_attn.o_proj",
-          "model.language_model.layers.9.self_attn.q_proj",
-          "model.language_model.layers.9.self_attn.v_proj"
-        ]
-      },
-      "group_mxfp8_shared_experts": {
-        "input_activations": {
-          "dynamic": true,
-          "num_bits": 8,
-          "type": "float",
-          "group_size": 32
-        },
-        "weights": {
-          "dynamic": false,
-          "num_bits": 8,
-          "type": "float",
-          "group_size": 32,
-          "weight_block_size": [
-            1,
-            32
-          ]
-        },
-        "targets": [
-          "model.language_model.layers.10.mlp.shared_experts.down_proj",
-          "model.language_model.layers.10.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.10.mlp.shared_experts.up_proj",
-          "model.language_model.layers.11.mlp.shared_experts.down_proj",
-          "model.language_model.layers.11.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.11.mlp.shared_experts.up_proj",
-          "model.language_model.layers.12.mlp.shared_experts.down_proj",
-          "model.language_model.layers.12.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.12.mlp.shared_experts.up_proj",
-          "model.language_model.layers.13.mlp.shared_experts.down_proj",
-          "model.language_model.layers.13.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.13.mlp.shared_experts.up_proj",
-          "model.language_model.layers.14.mlp.shared_experts.down_proj",
-          "model.language_model.layers.14.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.14.mlp.shared_experts.up_proj",
-          "model.language_model.layers.15.mlp.shared_experts.down_proj",
-          "model.language_model.layers.15.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.15.mlp.shared_experts.up_proj",
-          "model.language_model.layers.16.mlp.shared_experts.down_proj",
-          "model.language_model.layers.16.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.16.mlp.shared_experts.up_proj",
-          "model.language_model.layers.17.mlp.shared_experts.down_proj",
-          "model.language_model.layers.17.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.17.mlp.shared_experts.up_proj",
-          "model.language_model.layers.18.mlp.shared_experts.down_proj",
-          "model.language_model.layers.18.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.18.mlp.shared_experts.up_proj",
-          "model.language_model.layers.19.mlp.shared_experts.down_proj",
-          "model.language_model.layers.19.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.19.mlp.shared_experts.up_proj",
-          "model.language_model.layers.20.mlp.shared_experts.down_proj",
-          "model.language_model.layers.20.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.20.mlp.shared_experts.up_proj",
-          "model.language_model.layers.21.mlp.shared_experts.down_proj",
-          "model.language_model.layers.21.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.21.mlp.shared_experts.up_proj",
-          "model.language_model.layers.22.mlp.shared_experts.down_proj",
-          "model.language_model.layers.22.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.22.mlp.shared_experts.up_proj",
-          "model.language_model.layers.23.mlp.shared_experts.down_proj",
-          "model.language_model.layers.23.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.23.mlp.shared_experts.up_proj",
-          "model.language_model.layers.24.mlp.shared_experts.down_proj",
-          "model.language_model.layers.24.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.24.mlp.shared_experts.up_proj",
-          "model.language_model.layers.25.mlp.shared_experts.down_proj",
-          "model.language_model.layers.25.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.25.mlp.shared_experts.up_proj",
-          "model.language_model.layers.26.mlp.shared_experts.down_proj",
-          "model.language_model.layers.26.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.26.mlp.shared_experts.up_proj",
-          "model.language_model.layers.27.mlp.shared_experts.down_proj",
-          "model.language_model.layers.27.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.27.mlp.shared_experts.up_proj",
-          "model.language_model.layers.28.mlp.shared_experts.down_proj",
-          "model.language_model.layers.28.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.28.mlp.shared_experts.up_proj",
-          "model.language_model.layers.29.mlp.shared_experts.down_proj",
-          "model.language_model.layers.29.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.29.mlp.shared_experts.up_proj",
-          "model.language_model.layers.3.mlp.shared_experts.down_proj",
-          "model.language_model.layers.3.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.3.mlp.shared_experts.up_proj",
-          "model.language_model.layers.30.mlp.shared_experts.down_proj",
-          "model.language_model.layers.30.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.30.mlp.shared_experts.up_proj",
-          "model.language_model.layers.31.mlp.shared_experts.down_proj",
-          "model.language_model.layers.31.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.31.mlp.shared_experts.up_proj",
-          "model.language_model.layers.32.mlp.shared_experts.down_proj",
-          "model.language_model.layers.32.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.32.mlp.shared_experts.up_proj",
-          "model.language_model.layers.33.mlp.shared_experts.down_proj",
-          "model.language_model.layers.33.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.33.mlp.shared_experts.up_proj",
-          "model.language_model.layers.34.mlp.shared_experts.down_proj",
-          "model.language_model.layers.34.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.34.mlp.shared_experts.up_proj",
-          "model.language_model.layers.35.mlp.shared_experts.down_proj",
-          "model.language_model.layers.35.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.35.mlp.shared_experts.up_proj",
-          "model.language_model.layers.36.mlp.shared_experts.down_proj",
-          "model.language_model.layers.36.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.36.mlp.shared_experts.up_proj",
-          "model.language_model.layers.37.mlp.shared_experts.down_proj",
-          "model.language_model.layers.37.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.37.mlp.shared_experts.up_proj",
-          "model.language_model.layers.38.mlp.shared_experts.down_proj",
-          "model.language_model.layers.38.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.38.mlp.shared_experts.up_proj",
-          "model.language_model.layers.39.mlp.shared_experts.down_proj",
-          "model.language_model.layers.39.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.39.mlp.shared_experts.up_proj",
-          "model.language_model.layers.4.mlp.shared_experts.down_proj",
-          "model.language_model.layers.4.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.4.mlp.shared_experts.up_proj",
-          "model.language_model.layers.40.mlp.shared_experts.down_proj",
-          "model.language_model.layers.40.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.40.mlp.shared_experts.up_proj",
-          "model.language_model.layers.41.mlp.shared_experts.down_proj",
-          "model.language_model.layers.41.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.41.mlp.shared_experts.up_proj",
-          "model.language_model.layers.42.mlp.shared_experts.down_proj",
-          "model.language_model.layers.42.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.42.mlp.shared_experts.up_proj",
-          "model.language_model.layers.43.mlp.shared_experts.down_proj",
-          "model.language_model.layers.43.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.43.mlp.shared_experts.up_proj",
-          "model.language_model.layers.44.mlp.shared_experts.down_proj",
-          "model.language_model.layers.44.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.44.mlp.shared_experts.up_proj",
-          "model.language_model.layers.45.mlp.shared_experts.down_proj",
-          "model.language_model.layers.45.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.45.mlp.shared_experts.up_proj",
-          "model.language_model.layers.5.mlp.shared_experts.down_proj",
-          "model.language_model.layers.5.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.5.mlp.shared_experts.up_proj",
-          "model.language_model.layers.6.mlp.shared_experts.down_proj",
-          "model.language_model.layers.6.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.6.mlp.shared_experts.up_proj",
-          "model.language_model.layers.7.mlp.shared_experts.down_proj",
-          "model.language_model.layers.7.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.7.mlp.shared_experts.up_proj",
-          "model.language_model.layers.8.mlp.shared_experts.down_proj",
-          "model.language_model.layers.8.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.8.mlp.shared_experts.up_proj",
-          "model.language_model.layers.9.mlp.shared_experts.down_proj",
-          "model.language_model.layers.9.mlp.shared_experts.gate_proj",
-          "model.language_model.layers.9.mlp.shared_experts.up_proj"
-        ]
-      },
-      "group_w4a16_nvfp4_mtp_routed_experts": {
-        "weights": {
-          "dynamic": false,
-          "num_bits": 4,
-          "type": "float",
-          "group_size": 16
-        },
-        "targets": [
-          "model.language_model.layers.45.mlp.experts"
-        ]
-      },
-      "group_vision_mxfp8": {
-        "targets": [
-          "model.visual.blocks.0.attn.proj",
-          "model.visual.blocks.0.attn.qkv",
-          "model.visual.blocks.1.attn.proj",
-          "model.visual.blocks.1.attn.qkv",
-          "model.visual.blocks.10.attn.proj",
-          "model.visual.blocks.10.attn.qkv",
-          "model.visual.blocks.11.attn.proj",
-          "model.visual.blocks.11.attn.qkv",
-          "model.visual.blocks.12.attn.proj",
-          "model.visual.blocks.12.attn.qkv",
-          "model.visual.blocks.13.attn.proj",
-          "model.visual.blocks.13.attn.qkv",
-          "model.visual.blocks.14.attn.proj",
-          "model.visual.blocks.14.attn.qkv",
-          "model.visual.blocks.15.attn.proj",
-          "model.visual.blocks.15.attn.qkv",
-          "model.visual.blocks.16.attn.proj",
-          "model.visual.blocks.16.attn.qkv",
-          "model.visual.blocks.17.attn.proj",
-          "model.visual.blocks.17.attn.qkv",
-          "model.visual.blocks.18.attn.proj",
-          "model.visual.blocks.18.attn.qkv",
-          "model.visual.blocks.19.attn.proj",
-          "model.visual.blocks.19.attn.qkv",
-          "model.visual.blocks.2.attn.proj",
-          "model.visual.blocks.2.attn.qkv",
-          "model.visual.blocks.20.attn.proj",
-          "model.visual.blocks.20.attn.qkv",
-          "model.visual.blocks.21.attn.proj",
-          "model.visual.blocks.21.attn.qkv",
-          "model.visual.blocks.22.attn.proj",
-          "model.visual.blocks.22.attn.qkv",
-          "model.visual.blocks.23.attn.proj",
-          "model.visual.blocks.23.attn.qkv",
-          "model.visual.blocks.3.attn.proj",
-          "model.visual.blocks.3.attn.qkv",
-          "model.visual.blocks.4.attn.proj",
-          "model.visual.blocks.4.attn.qkv",
-          "model.visual.blocks.5.attn.proj",
-          "model.visual.blocks.5.attn.qkv",
-          "model.visual.blocks.6.attn.proj",
-          "model.visual.blocks.6.attn.qkv",
-          "model.visual.blocks.7.attn.proj",
-          "model.visual.blocks.7.attn.qkv",
-          "model.visual.blocks.8.attn.proj",
-          "model.visual.blocks.8.attn.qkv",
-          "model.visual.blocks.9.attn.proj",
-          "model.visual.blocks.9.attn.qkv"
-        ],
-        "weights": {
-          "dynamic": false,
-          "num_bits": 8,
-          "type": "float",
-          "group_size": 32,
-          "weight_block_size": [
-            1,
-            32
-          ]
-        },
-        "input_activations": {
-          "dynamic": true,
-          "num_bits": 8,
-          "type": "float",
-          "group_size": 32
-        }
-      },
-      "group_vision_w4a16_nvfp4": {
-        "targets": [
-          "model.visual.blocks.0.mlp.down_proj",
-          "model.visual.blocks.0.mlp.gate_proj",
-          "model.visual.blocks.0.mlp.up_proj",
-          "model.visual.blocks.1.mlp.down_proj",
-          "model.visual.blocks.1.mlp.gate_proj",
-          "model.visual.blocks.1.mlp.up_proj",
-          "model.visual.blocks.10.mlp.down_proj",
-          "model.visual.blocks.10.mlp.gate_proj",
-          "model.visual.blocks.10.mlp.up_proj",
-          "model.visual.blocks.11.mlp.down_proj",
-          "model.visual.blocks.11.mlp.gate_proj",
-          "model.visual.blocks.11.mlp.up_proj",
-          "model.visual.blocks.12.mlp.down_proj",
-          "model.visual.blocks.12.mlp.gate_proj",
-          "model.visual.blocks.12.mlp.up_proj",
-          "model.visual.blocks.13.mlp.down_proj",
-          "model.visual.blocks.13.mlp.gate_proj",
-          "model.visual.blocks.13.mlp.up_proj",
-          "model.visual.blocks.14.mlp.down_proj",
-          "model.visual.blocks.14.mlp.gate_proj",
-          "model.visual.blocks.14.mlp.up_proj",
-          "model.visual.blocks.15.mlp.down_proj",
-          "model.visual.blocks.15.mlp.gate_proj",
-          "model.visual.blocks.15.mlp.up_proj",
-          "model.visual.blocks.16.mlp.down_proj",
-          "model.visual.blocks.16.mlp.gate_proj",
-          "model.visual.blocks.16.mlp.up_proj",
-          "model.visual.blocks.17.mlp.down_proj",
-          "model.visual.blocks.17.mlp.gate_proj",
-          "model.visual.blocks.17.mlp.up_proj",
-          "model.visual.blocks.18.mlp.down_proj",
-          "model.visual.blocks.18.mlp.gate_proj",
-          "model.visual.blocks.18.mlp.up_proj",
-          "model.visual.blocks.19.mlp.down_proj",
-          "model.visual.blocks.19.mlp.gate_proj",
-          "model.visual.blocks.19.mlp.up_proj",
-          "model.visual.blocks.2.mlp.down_proj",
-          "model.visual.blocks.2.mlp.gate_proj",
-          "model.visual.blocks.2.mlp.up_proj",
-          "model.visual.blocks.20.mlp.down_proj",
-          "model.visual.blocks.20.mlp.gate_proj",
-          "model.visual.blocks.20.mlp.up_proj",
-          "model.visual.blocks.21.mlp.down_proj",
-          "model.visual.blocks.21.mlp.gate_proj",
-          "model.visual.blocks.21.mlp.up_proj",
-          "model.visual.blocks.22.mlp.down_proj",
-          "model.visual.blocks.22.mlp.gate_proj",
-          "model.visual.blocks.22.mlp.up_proj",
-          "model.visual.blocks.23.mlp.down_proj",
-          "model.visual.blocks.23.mlp.gate_proj",
-          "model.visual.blocks.23.mlp.up_proj",
-          "model.visual.blocks.3.mlp.down_proj",
-          "model.visual.blocks.3.mlp.gate_proj",
-          "model.visual.blocks.3.mlp.up_proj",
-          "model.visual.blocks.4.mlp.down_proj",
-          "model.visual.blocks.4.mlp.gate_proj",
-          "model.visual.blocks.4.mlp.up_proj",
-          "model.visual.blocks.5.mlp.down_proj",
-          "model.visual.blocks.5.mlp.gate_proj",
-          "model.visual.blocks.5.mlp.up_proj",
-          "model.visual.blocks.6.mlp.down_proj",
-          "model.visual.blocks.6.mlp.gate_proj",
-          "model.visual.blocks.6.mlp.up_proj",
-          "model.visual.blocks.7.mlp.down_proj",
-          "model.visual.blocks.7.mlp.gate_proj",
-          "model.visual.blocks.7.mlp.up_proj",
-          "model.visual.blocks.8.mlp.down_proj",
-          "model.visual.blocks.8.mlp.gate_proj",
-          "model.visual.blocks.8.mlp.up_proj",
-          "model.visual.blocks.9.mlp.down_proj",
-          "model.visual.blocks.9.mlp.gate_proj",
-          "model.visual.blocks.9.mlp.up_proj",
-          "model.visual.merger.down_proj",
-          "model.visual.merger.gate_proj",
-          "model.visual.merger.proj",
-          "model.visual.merger.up_proj"
-        ],
-        "weights": {
-          "dynamic": false,
-          "num_bits": 4,
-          "type": "float",
-          "group_size": 16
-        }
-      }
-    },
-    "ignore": [],
-    "producer": {
-      "name": "modelopt",
-      "version": "0.39.0.dev290+gf9d9a71de.d20260407"
-    },
+    "exclude_modules": [],
     "quant_algo": "MIXED_PRECISION",
     "quant_method": "modelopt",
     "quantized_layers": {
-      "model.language_model.layers.10.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.11.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.12.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.13.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.14.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.15.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.16.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.17.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.18.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.19.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.20.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.21.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.22.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.23.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.24.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.25.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.26.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.27.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.28.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.29.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.3.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.30.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.31.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.32.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.33.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.34.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.35.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.36.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.37.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.38.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.39.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.4.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.40.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.41.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.42.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.43.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.44.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.5.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.6.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.7.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.8.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.9.mlp.experts": {
-        "group_size": 16,
-        "quant_algo": "NVFP4",
-        "weight_scale_encoding": "csf"
-      },
-      "model.language_model.layers.0.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.0.self_attn.f_a_proj": {
+      "lm_head": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.0.self_attn.f_b_proj": {
+      "model.language_model.layers.0.mlp.down_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.0.self_attn.g_a_proj": {
+      "model.language_model.layers.0.mlp.gate_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.0.self_attn.g_b_proj": {
+      "model.language_model.layers.0.mlp.up_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.0.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.0.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.0.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.0.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.1.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.1.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.1.self_attn.f_b_proj": {
+      "model.language_model.layers.1.mlp.down_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.1.self_attn.g_a_proj": {
+      "model.language_model.layers.1.mlp.gate_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.1.self_attn.g_b_proj": {
+      "model.language_model.layers.1.mlp.up_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.1.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.1.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.1.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.1.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
+      },
+      "model.language_model.layers.10.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
       "model.language_model.layers.10.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.10.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.10.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.10.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.10.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.10.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.10.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.10.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.10.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.10.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.10.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.10.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
+      },
+      "model.language_model.layers.11.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
       "model.language_model.layers.11.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.11.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.11.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.11.self_attn.indexer.weights_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.11.self_attn.indexer.wk": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.11.self_attn.indexer.wq_b": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.11.self_attn.kv_a_proj_with_mqa": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.11.self_attn.kv_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.11.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.11.self_attn.q_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.11.self_attn.q_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
+      },
+      "model.language_model.layers.12.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
       "model.language_model.layers.12.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.12.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.12.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.12.self_attn.b_proj": {
+      "model.language_model.layers.12.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.12.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.12.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.12.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.12.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.12.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.12.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.12.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
       "model.language_model.layers.12.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
+      },
+      "model.language_model.layers.13.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
       "model.language_model.layers.13.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.13.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
       "model.language_model.layers.13.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.13.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.13.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.13.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.13.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.13.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.13.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.13.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.13.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.13.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.14.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.15.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.15.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.15.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.15.self_attn.indexer.weights_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.15.self_attn.indexer.wk": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.15.self_attn.indexer.wq_b": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.15.self_attn.kv_a_proj_with_mqa": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.15.self_attn.kv_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.15.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.15.self_attn.q_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.15.self_attn.q_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.16.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.17.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.18.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.19.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.19.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.19.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.19.self_attn.indexer.weights_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.19.self_attn.indexer.wk": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.19.self_attn.indexer.wq_b": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.19.self_attn.kv_a_proj_with_mqa": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.19.self_attn.kv_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.19.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.19.self_attn.q_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.19.self_attn.q_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.2.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.2.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.2.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.2.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.2.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.2.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.2.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.2.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.2.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.20.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.21.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.22.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.23.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.23.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.23.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.23.self_attn.indexer.weights_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.23.self_attn.indexer.wk": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.23.self_attn.indexer.wq_b": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.23.self_attn.kv_a_proj_with_mqa": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.23.self_attn.kv_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.23.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.23.self_attn.q_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.23.self_attn.q_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.24.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.25.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.26.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.27.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.27.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.27.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.27.self_attn.indexer.weights_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.27.self_attn.indexer.wk": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.27.self_attn.indexer.wq_b": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.27.self_attn.kv_a_proj_with_mqa": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.27.self_attn.kv_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.27.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.27.self_attn.q_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.27.self_attn.q_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.28.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.29.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.29.mlp.shared_experts.gate_proj": {
+      "model.language_model.layers.13.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.29.mlp.shared_experts.up_proj": {
+      "model.language_model.layers.13.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.29.self_attn.b_proj": {
+      "model.language_model.layers.13.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.29.self_attn.f_a_proj": {
+      "model.language_model.layers.13.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.29.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.14.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.29.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.14.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.29.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.14.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.29.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.14.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.29.self_attn.o_proj": {
+      "model.language_model.layers.14.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.29.self_attn.q_proj": {
+      "model.language_model.layers.14.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.29.self_attn.v_proj": {
+      "model.language_model.layers.14.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.3.mlp.shared_experts.down_proj": {
+      "model.language_model.layers.14.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.3.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.15.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.3.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.15.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.3.self_attn.indexer.weights_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.15.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.3.self_attn.indexer.wk": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.15.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.3.self_attn.indexer.wq_b": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.15.self_attn.kv_a_proj_with_mqa": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.3.self_attn.kv_a_proj_with_mqa": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.15.self_attn.o_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.3.self_attn.kv_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.15.self_attn.q_a_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.3.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.15.self_attn.q_b_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.3.self_attn.q_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.16.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.3.self_attn.q_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.16.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.30.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.16.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.30.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.16.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.30.mlp.shared_experts.up_proj": {
+      "model.language_model.layers.16.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.30.self_attn.b_proj": {
+      "model.language_model.layers.16.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.30.self_attn.f_a_proj": {
+      "model.language_model.layers.16.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.30.self_attn.f_b_proj": {
+      "model.language_model.layers.16.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.30.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.17.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.30.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.17.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.30.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.17.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.30.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.17.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.30.self_attn.q_proj": {
+      "model.language_model.layers.17.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.30.self_attn.v_proj": {
+      "model.language_model.layers.17.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.31.mlp.shared_experts.down_proj": {
+      "model.language_model.layers.17.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.31.mlp.shared_experts.gate_proj": {
+      "model.language_model.layers.17.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.31.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.18.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.31.self_attn.indexer.weights_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.18.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.31.self_attn.indexer.wk": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.18.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.31.self_attn.indexer.wq_b": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.18.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.31.self_attn.kv_a_proj_with_mqa": {
+      "model.language_model.layers.18.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.31.self_attn.kv_b_proj": {
+      "model.language_model.layers.18.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.31.self_attn.o_proj": {
+      "model.language_model.layers.18.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.31.self_attn.q_a_proj": {
+      "model.language_model.layers.18.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.31.self_attn.q_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.19.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.32.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.19.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.32.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.19.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.32.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.19.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.32.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.19.self_attn.kv_a_proj_with_mqa": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.32.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.19.self_attn.o_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.32.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.19.self_attn.q_a_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.32.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.19.self_attn.q_b_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.32.self_attn.g_b_proj": {
+      "model.language_model.layers.2.mlp.down_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.32.self_attn.k_proj": {
+      "model.language_model.layers.2.mlp.gate_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.32.self_attn.o_proj": {
+      "model.language_model.layers.2.mlp.up_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.32.self_attn.q_proj": {
+      "model.language_model.layers.2.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.32.self_attn.v_proj": {
+      "model.language_model.layers.2.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.33.mlp.shared_experts.down_proj": {
+      "model.language_model.layers.2.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.33.mlp.shared_experts.gate_proj": {
+      "model.language_model.layers.2.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.33.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.20.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.33.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.20.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.33.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.20.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.33.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.20.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.33.self_attn.g_a_proj": {
+      "model.language_model.layers.20.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.33.self_attn.g_b_proj": {
+      "model.language_model.layers.20.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.33.self_attn.k_proj": {
+      "model.language_model.layers.20.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.33.self_attn.o_proj": {
+      "model.language_model.layers.20.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.33.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.21.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.33.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.21.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.34.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.21.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.34.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.21.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.34.mlp.shared_experts.up_proj": {
+      "model.language_model.layers.21.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.34.self_attn.b_proj": {
+      "model.language_model.layers.21.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.34.self_attn.f_a_proj": {
+      "model.language_model.layers.21.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.34.self_attn.f_b_proj": {
+      "model.language_model.layers.21.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.34.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.22.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.34.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.22.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.34.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.22.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.34.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.22.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.34.self_attn.q_proj": {
+      "model.language_model.layers.22.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.34.self_attn.v_proj": {
+      "model.language_model.layers.22.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.35.mlp.shared_experts.down_proj": {
+      "model.language_model.layers.22.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.35.mlp.shared_experts.gate_proj": {
+      "model.language_model.layers.22.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.35.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.23.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.35.self_attn.indexer.weights_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.23.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.35.self_attn.indexer.wk": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.23.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.35.self_attn.indexer.wq_b": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.23.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.35.self_attn.kv_a_proj_with_mqa": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.23.self_attn.kv_a_proj_with_mqa": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.35.self_attn.kv_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.23.self_attn.o_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.35.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.23.self_attn.q_a_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.35.self_attn.q_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.23.self_attn.q_b_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.35.self_attn.q_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.24.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.36.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.24.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.36.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.24.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.36.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.24.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.36.self_attn.b_proj": {
+      "model.language_model.layers.24.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.36.self_attn.f_a_proj": {
+      "model.language_model.layers.24.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.36.self_attn.f_b_proj": {
+      "model.language_model.layers.24.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.36.self_attn.g_a_proj": {
+      "model.language_model.layers.24.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.36.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.25.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.36.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.25.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.36.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.25.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.36.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.25.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.36.self_attn.v_proj": {
+      "model.language_model.layers.25.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.37.mlp.shared_experts.down_proj": {
+      "model.language_model.layers.25.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.37.mlp.shared_experts.gate_proj": {
+      "model.language_model.layers.25.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.37.mlp.shared_experts.up_proj": {
+      "model.language_model.layers.25.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.37.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.26.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.37.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.26.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.37.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.26.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.37.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.26.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.37.self_attn.g_b_proj": {
+      "model.language_model.layers.26.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.37.self_attn.k_proj": {
+      "model.language_model.layers.26.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.37.self_attn.o_proj": {
+      "model.language_model.layers.26.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.37.self_attn.q_proj": {
+      "model.language_model.layers.26.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.37.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.27.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.38.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.27.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.38.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.27.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.38.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.27.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.38.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.27.self_attn.kv_a_proj_with_mqa": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.38.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.27.self_attn.o_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.38.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.27.self_attn.q_a_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.38.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.27.self_attn.q_b_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.38.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.28.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.38.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.28.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.38.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.28.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.38.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.28.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.38.self_attn.v_proj": {
+      "model.language_model.layers.28.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.39.mlp.shared_experts.down_proj": {
+      "model.language_model.layers.28.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.39.mlp.shared_experts.gate_proj": {
+      "model.language_model.layers.28.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.39.mlp.shared_experts.up_proj": {
+      "model.language_model.layers.28.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.39.self_attn.indexer.weights_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.29.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.39.self_attn.indexer.wk": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.29.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.39.self_attn.indexer.wq_b": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.29.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.39.self_attn.kv_a_proj_with_mqa": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.29.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.39.self_attn.kv_b_proj": {
+      "model.language_model.layers.29.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.39.self_attn.o_proj": {
+      "model.language_model.layers.29.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.39.self_attn.q_a_proj": {
+      "model.language_model.layers.29.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.39.self_attn.q_b_proj": {
+      "model.language_model.layers.29.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.4.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.3.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.4.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.3.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.4.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.3.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.4.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.3.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.4.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.3.self_attn.kv_a_proj_with_mqa": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.4.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.3.self_attn.o_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.4.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.3.self_attn.q_a_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.4.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.3.self_attn.q_b_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.4.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.30.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.4.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.30.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.4.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.30.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.4.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.30.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.40.mlp.shared_experts.down_proj": {
+      "model.language_model.layers.30.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.40.mlp.shared_experts.gate_proj": {
+      "model.language_model.layers.30.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.40.mlp.shared_experts.up_proj": {
+      "model.language_model.layers.30.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.40.self_attn.b_proj": {
+      "model.language_model.layers.30.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.40.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.31.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.40.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.31.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.40.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.31.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.40.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.31.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.40.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.31.self_attn.kv_a_proj_with_mqa": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.40.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.31.self_attn.o_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.40.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.31.self_attn.q_a_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.40.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.31.self_attn.q_b_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.41.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.32.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.41.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.32.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.41.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.32.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.41.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.32.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.41.self_attn.f_a_proj": {
+      "model.language_model.layers.32.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.41.self_attn.f_b_proj": {
+      "model.language_model.layers.32.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.41.self_attn.g_a_proj": {
+      "model.language_model.layers.32.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.41.self_attn.g_b_proj": {
+      "model.language_model.layers.32.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.41.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.33.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.41.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.33.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.41.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.33.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.41.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.33.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.42.mlp.shared_experts.down_proj": {
+      "model.language_model.layers.33.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.42.mlp.shared_experts.gate_proj": {
+      "model.language_model.layers.33.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.42.mlp.shared_experts.up_proj": {
+      "model.language_model.layers.33.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.42.self_attn.b_proj": {
+      "model.language_model.layers.33.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.42.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.34.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.42.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.34.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.42.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.34.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.42.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.34.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.42.self_attn.k_proj": {
+      "model.language_model.layers.34.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.42.self_attn.o_proj": {
+      "model.language_model.layers.34.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.42.self_attn.q_proj": {
+      "model.language_model.layers.34.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.42.self_attn.v_proj": {
+      "model.language_model.layers.34.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.43.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.35.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.43.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.35.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.43.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.35.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.43.self_attn.indexer.weights_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.35.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.43.self_attn.indexer.wk": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.35.self_attn.kv_a_proj_with_mqa": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.43.self_attn.indexer.wq_b": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.35.self_attn.o_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.43.self_attn.kv_a_proj_with_mqa": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.35.self_attn.q_a_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.43.self_attn.kv_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.35.self_attn.q_b_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.43.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.36.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.43.self_attn.q_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.36.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.43.self_attn.q_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.36.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.44.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.36.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.44.mlp.shared_experts.gate_proj": {
+      "model.language_model.layers.36.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.44.mlp.shared_experts.up_proj": {
+      "model.language_model.layers.36.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.44.self_attn.b_proj": {
+      "model.language_model.layers.36.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.44.self_attn.f_a_proj": {
+      "model.language_model.layers.36.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.44.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.37.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.44.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.37.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.44.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.37.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.44.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.37.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.44.self_attn.o_proj": {
+      "model.language_model.layers.37.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.44.self_attn.q_proj": {
+      "model.language_model.layers.37.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.44.self_attn.v_proj": {
+      "model.language_model.layers.37.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
-      },
-      "model.language_model.layers.45.mlp.experts": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.45.mlp.shared_experts.down_proj": {
+      "model.language_model.layers.37.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.45.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.38.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.45.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.38.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.45.self_attn.indexer.weights_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.38.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.45.self_attn.indexer.wk": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.38.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.45.self_attn.indexer.wq_b": {
+      "model.language_model.layers.38.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.45.self_attn.kv_a_proj_with_mqa": {
+      "model.language_model.layers.38.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.45.self_attn.kv_b_proj": {
+      "model.language_model.layers.38.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.45.self_attn.o_proj": {
+      "model.language_model.layers.38.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.45.self_attn.q_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.39.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.45.self_attn.q_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.39.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.5.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.39.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.5.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.39.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.5.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.39.self_attn.kv_a_proj_with_mqa": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.5.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.39.self_attn.o_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.5.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.39.self_attn.q_a_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.5.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.39.self_attn.q_b_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.5.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.4.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.5.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.4.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.5.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.4.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.5.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.4.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.5.self_attn.q_proj": {
+      "model.language_model.layers.4.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.5.self_attn.v_proj": {
+      "model.language_model.layers.4.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.6.mlp.shared_experts.down_proj": {
+      "model.language_model.layers.4.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.6.mlp.shared_experts.gate_proj": {
+      "model.language_model.layers.4.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.6.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.40.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.6.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.40.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.6.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.40.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.6.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.40.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.6.self_attn.g_a_proj": {
+      "model.language_model.layers.40.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.6.self_attn.g_b_proj": {
+      "model.language_model.layers.40.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.6.self_attn.k_proj": {
+      "model.language_model.layers.40.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.6.self_attn.o_proj": {
+      "model.language_model.layers.40.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.6.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.41.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.6.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.41.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.7.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.41.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.7.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.41.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.7.mlp.shared_experts.up_proj": {
+      "model.language_model.layers.41.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.7.self_attn.indexer.weights_proj": {
+      "model.language_model.layers.41.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.7.self_attn.indexer.wk": {
+      "model.language_model.layers.41.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.7.self_attn.indexer.wq_b": {
+      "model.language_model.layers.41.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.7.self_attn.kv_a_proj_with_mqa": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.42.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.7.self_attn.kv_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.42.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.7.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.42.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.7.self_attn.q_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.42.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.7.self_attn.q_b_proj": {
+      "model.language_model.layers.42.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.8.mlp.shared_experts.down_proj": {
+      "model.language_model.layers.42.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.8.mlp.shared_experts.gate_proj": {
+      "model.language_model.layers.42.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.8.mlp.shared_experts.up_proj": {
+      "model.language_model.layers.42.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.8.self_attn.b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.43.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.8.self_attn.f_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.43.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.8.self_attn.f_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.43.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.8.self_attn.g_a_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.43.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.8.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.43.self_attn.kv_a_proj_with_mqa": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.8.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.43.self_attn.o_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.8.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.43.self_attn.q_a_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.8.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.43.self_attn.q_b_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.8.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.44.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.language_model.layers.9.mlp.shared_experts.down_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.44.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.9.mlp.shared_experts.gate_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.44.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.9.mlp.shared_experts.up_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.44.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.language_model.layers.9.self_attn.b_proj": {
+      "model.language_model.layers.44.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.9.self_attn.f_a_proj": {
+      "model.language_model.layers.44.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.9.self_attn.f_b_proj": {
+      "model.language_model.layers.44.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.9.self_attn.g_a_proj": {
+      "model.language_model.layers.44.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.language_model.layers.9.self_attn.g_b_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
-      "model.language_model.layers.9.self_attn.k_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.mlp.shared_experts.down_proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
-      "model.language_model.layers.9.self_attn.o_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.mlp.shared_experts.gate_proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
-      "model.language_model.layers.9.self_attn.q_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.mlp.shared_experts.up_proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
-      "model.language_model.layers.9.self_attn.v_proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.self_attn.indexer.weights_proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
-      "model.visual.blocks.0.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.self_attn.indexer.wk": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
-      "model.visual.blocks.0.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.self_attn.indexer.wq_b": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
-      "model.visual.blocks.1.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.self_attn.kv_a_proj_with_mqa": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
-      "model.visual.blocks.1.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.self_attn.kv_b_proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
-      "model.visual.blocks.10.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.self_attn.o_proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
-      "model.visual.blocks.10.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.self_attn.q_a_proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
-      "model.visual.blocks.11.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.45.self_attn.q_b_proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
-      "model.visual.blocks.11.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.5.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.visual.blocks.12.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.5.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.12.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.5.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.13.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.5.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.13.attn.qkv": {
+      "model.language_model.layers.5.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.14.attn.proj": {
+      "model.language_model.layers.5.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.14.attn.qkv": {
+      "model.language_model.layers.5.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.15.attn.proj": {
+      "model.language_model.layers.5.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.15.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.6.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.visual.blocks.16.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.6.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.16.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.6.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.17.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.6.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.17.attn.qkv": {
+      "model.language_model.layers.6.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.18.attn.proj": {
+      "model.language_model.layers.6.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.18.attn.qkv": {
+      "model.language_model.layers.6.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.19.attn.proj": {
+      "model.language_model.layers.6.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.19.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.7.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.visual.blocks.2.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.7.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.2.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.7.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.20.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.7.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.20.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.7.self_attn.kv_a_proj_with_mqa": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.21.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.7.self_attn.o_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.21.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.7.self_attn.q_a_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.22.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.7.self_attn.q_b_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.22.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.8.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.visual.blocks.23.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.8.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.23.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.8.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.3.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.8.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.3.attn.qkv": {
+      "model.language_model.layers.8.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.4.attn.proj": {
+      "model.language_model.layers.8.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.4.attn.qkv": {
+      "model.language_model.layers.8.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.5.attn.proj": {
+      "model.language_model.layers.8.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.5.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.9.mlp.experts": {
+        "group_size": 16,
+        "quant_algo": "NVFP4",
+        "weight_scale_encoding": "csf_32x2_v1"
       },
-      "model.visual.blocks.6.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.9.mlp.shared_experts.down_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.6.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.9.mlp.shared_experts.gate_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.7.attn.proj": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.language_model.layers.9.mlp.shared_experts.up_proj": {
+        "quant_algo": "FP8_PB_WO",
+        "weight_block_size": [
+          128,
+          128
+        ]
       },
-      "model.visual.blocks.7.attn.qkv": {
+      "model.language_model.layers.9.self_attn.k_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.8.attn.proj": {
+      "model.language_model.layers.9.self_attn.o_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.8.attn.qkv": {
+      "model.language_model.layers.9.self_attn.q_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.9.attn.proj": {
+      "model.language_model.layers.9.self_attn.v_proj": {
         "quant_algo": "MXFP8",
-        "group_size": 32
+        "weight_block_size": [
+          1,
+          32
+        ]
       },
-      "model.visual.blocks.9.attn.qkv": {
-        "quant_algo": "MXFP8",
-        "group_size": 32
+      "model.visual.blocks.0.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.0.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.0.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.0.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.0.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.1.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.1.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.1.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.1.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.1.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.10.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.10.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.10.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.10.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.10.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.11.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.11.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.11.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.11.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.11.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.12.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.12.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.12.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.12.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.12.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.13.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.13.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.13.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.13.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.13.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.14.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.14.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.14.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.14.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.14.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.15.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.15.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.15.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.15.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.15.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.16.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.16.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.16.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.16.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.16.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.17.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.17.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.17.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.17.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.17.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.18.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.18.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.18.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.18.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.18.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.19.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.19.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.19.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.19.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.19.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.2.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.2.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.2.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.2.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.2.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.20.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.20.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.20.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.20.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.20.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.21.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.21.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.21.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.21.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.21.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.22.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.22.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.22.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.22.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.22.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.23.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.23.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.23.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.23.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.23.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.3.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.3.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.3.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.3.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.3.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.4.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.4.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.4.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.4.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.4.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.5.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.5.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.5.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.5.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.5.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.6.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.6.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.6.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.6.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.6.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.7.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.7.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.7.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.7.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.7.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.8.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.8.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.8.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.8.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.8.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
+      },
+      "model.visual.blocks.9.attn.proj": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
+      },
+      "model.visual.blocks.9.attn.qkv": {
+        "group_size": 32,
+        "quant_algo": "MXFP8"
       },
       "model.visual.blocks.9.mlp.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.9.mlp.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.blocks.9.mlp.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.merger.down_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.merger.gate_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.merger.proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       },
       "model.visual.merger.up_proj": {
-        "quant_algo": "W4A16_NVFP4",
-        "group_size": 16
+        "group_size": 16,
+        "quant_algo": "W4A16_NVFP4"
       }
     }
   },
-  "return_dict": true,
   "text_config": {
-    "_name_or_path": "",
     "attention_bias": false,
     "attention_dropout": 0.0,
-    "bos_token_id": null,
-    "chunk_size_feed_forward": 0,
     "dtype": "bfloat16",
     "eos_token_id": [
       154820,
@@ -3682,10 +2995,6 @@
     "head_dim": 0,
     "hidden_act": "silu",
     "hidden_size": 4096,
-    "id2label": {
-      "0": "LABEL_0",
-      "1": "LABEL_1"
-    },
     "index_head_dim": 128,
     "index_kpool": 4,
     "index_kpool_always_select_tail": true,
@@ -3743,12 +3052,7 @@
     ],
     "initializer_range": 0.02,
     "intermediate_size": 12288,
-    "is_encoder_decoder": false,
     "kv_lora_rank": 512,
-    "label2id": {
-      "LABEL_0": 0,
-      "LABEL_1": 1
-    },
     "layer_types": [
       "linear_attention",
       "linear_attention",
@@ -3851,10 +3155,6 @@
       "num_heads": 64,
       "short_conv_kernel_size": 4
     },
-    "linear_conv_kernel_dim": 4,
-    "linear_head_dim": 128,
-    "linear_lower_bound": -5.0,
-    "linear_num_heads": 64,
     "max_position_embeddings": 1048576,
     "mhc": true,
     "mla_use_nope": true,
@@ -3917,16 +3217,12 @@
     "num_hidden_layers": 45,
     "num_key_value_heads": 64,
     "num_nextn_predict_layers": 1,
-    "output_attentions": false,
-    "output_hidden_states": false,
     "output_router_logits": false,
     "pad_token_id": 154820,
-    "problem_type": null,
     "q_lora_rank": 1536,
     "qk_head_dim": 256,
     "qk_nope_head_dim": 256,
     "qk_rope_head_dim": 0,
-    "return_dict": true,
     "rms_norm_eps": 1e-05,
     "routed_scaling_factor": 2.5,
     "router_aux_loss_coef": 0.001,
@@ -3940,42 +3236,25 @@
     "vocab_size": 154880
   },
   "tie_word_embeddings": false,
-  "transformers_version": "5.16.1",
+  "transformers_version": "5.16.0",
   "video_end_token_id": 154833,
   "video_start_token_id": 154832,
   "video_token_id": 154855,
   "vision_config": {
-    "_name_or_path": "",
-    "architectures": null,
     "attention_bias": true,
     "attention_dropout": 0.0,
-    "chunk_size_feed_forward": 0,
     "depth": 24,
-    "dtype": "bfloat16",
     "hidden_act": "silu",
     "hidden_size": 1024,
-    "id2label": {
-      "0": "LABEL_0",
-      "1": "LABEL_1"
-    },
     "image_size": 448,
     "in_channels": 3,
     "initializer_range": 0.02,
     "intermediate_size": 4096,
-    "is_encoder_decoder": false,
-    "label2id": {
-      "LABEL_0": 0,
-      "LABEL_1": 1
-    },
     "model_type": "glm5_next_vision",
     "num_heads": 16,
     "out_hidden_size": 4096,
-    "output_attentions": false,
-    "output_hidden_states": false,
     "patch_size": 14,
-    "problem_type": null,
     "projection_intermediate_size": 10240,
-    "return_dict": true,
     "rms_norm_eps": 1e-05,
     "spatial_merge_size": 2,
     "swiglu_limit": 10.0,
diff --git a/hf_quant_config.json b/hf_quant_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..49cbc8867168a8a58f7c1eeb75ea2d22c3481f62
--- /dev/null
+++ b/hf_quant_config.json
@@ -0,0 +1,2973 @@
+{
+  "exclude_modules": [],
+  "quant_algo": "MIXED_PRECISION",
+  "quant_method": "modelopt",
+  "quantized_layers": {
+    "lm_head": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.0.mlp.down_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.0.mlp.gate_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.0.mlp.up_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.0.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.0.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.0.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.0.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.1.mlp.down_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.1.mlp.gate_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.1.mlp.up_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.1.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.1.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.1.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.1.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.10.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.10.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.10.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.10.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.10.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.10.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.10.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.10.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.11.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.11.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.11.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.11.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.11.self_attn.kv_a_proj_with_mqa": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.11.self_attn.o_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.11.self_attn.q_a_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.11.self_attn.q_b_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.12.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.12.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.12.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.12.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.12.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.12.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.12.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.12.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.13.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.13.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.13.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.13.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.13.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.13.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.13.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.13.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.14.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.14.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.14.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.14.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.14.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.14.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.14.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.14.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.15.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.15.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.15.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.15.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.15.self_attn.kv_a_proj_with_mqa": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.15.self_attn.o_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.15.self_attn.q_a_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.15.self_attn.q_b_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.16.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.16.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.16.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.16.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.16.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.16.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.16.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.16.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.17.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.17.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.17.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.17.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.17.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.17.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.17.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.17.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.18.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.18.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.18.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.18.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.18.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.18.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.18.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.18.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.19.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.19.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.19.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.19.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.19.self_attn.kv_a_proj_with_mqa": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.19.self_attn.o_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.19.self_attn.q_a_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.19.self_attn.q_b_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.2.mlp.down_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.2.mlp.gate_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.2.mlp.up_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.2.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.2.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.2.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.2.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.20.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.20.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.20.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.20.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.20.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.20.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.20.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.20.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.21.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.21.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.21.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.21.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.21.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.21.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.21.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.21.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.22.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.22.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.22.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.22.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.22.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.22.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.22.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.22.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.23.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.23.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.23.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.23.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.23.self_attn.kv_a_proj_with_mqa": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.23.self_attn.o_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.23.self_attn.q_a_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.23.self_attn.q_b_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.24.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.24.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.24.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.24.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.24.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.24.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.24.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.24.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.25.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.25.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.25.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.25.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.25.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.25.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.25.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.25.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.26.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.26.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.26.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.26.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.26.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.26.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.26.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.26.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.27.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.27.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.27.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.27.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.27.self_attn.kv_a_proj_with_mqa": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.27.self_attn.o_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.27.self_attn.q_a_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.27.self_attn.q_b_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.28.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.28.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.28.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.28.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.28.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.28.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.28.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.28.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.29.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.29.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.29.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.29.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.29.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.29.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.29.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.29.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.3.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.3.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.3.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.3.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.3.self_attn.kv_a_proj_with_mqa": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.3.self_attn.o_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.3.self_attn.q_a_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.3.self_attn.q_b_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.30.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.30.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.30.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.30.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.30.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.30.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.30.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.30.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.31.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.31.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.31.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.31.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.31.self_attn.kv_a_proj_with_mqa": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.31.self_attn.o_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.31.self_attn.q_a_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.31.self_attn.q_b_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.32.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.32.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.32.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.32.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.32.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.32.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.32.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.32.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.33.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.33.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.33.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.33.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.33.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.33.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.33.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.33.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.34.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.34.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.34.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.34.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.34.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.34.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.34.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.34.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.35.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.35.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.35.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.35.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.35.self_attn.kv_a_proj_with_mqa": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.35.self_attn.o_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.35.self_attn.q_a_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.35.self_attn.q_b_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.36.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.36.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.36.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.36.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.36.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.36.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.36.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.36.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.37.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.37.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.37.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.37.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.37.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.37.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.37.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.37.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.38.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.38.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.38.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.38.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.38.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.38.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.38.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.38.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.39.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.39.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.39.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.39.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.39.self_attn.kv_a_proj_with_mqa": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.39.self_attn.o_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.39.self_attn.q_a_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.39.self_attn.q_b_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.4.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.4.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.4.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.4.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.4.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.4.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.4.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.4.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.40.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.40.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.40.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.40.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.40.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.40.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.40.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.40.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.41.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.41.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.41.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.41.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.41.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.41.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.41.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.41.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.42.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.42.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.42.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.42.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.42.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.42.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.42.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.42.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.43.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.43.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.43.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.43.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.43.self_attn.kv_a_proj_with_mqa": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.43.self_attn.o_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.43.self_attn.q_a_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.43.self_attn.q_b_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.44.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.44.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.44.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.44.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.44.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.44.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.44.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.44.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.45.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.language_model.layers.45.mlp.shared_experts.down_proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.language_model.layers.45.mlp.shared_experts.gate_proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.language_model.layers.45.mlp.shared_experts.up_proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.language_model.layers.45.self_attn.indexer.weights_proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.language_model.layers.45.self_attn.indexer.wk": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.language_model.layers.45.self_attn.indexer.wq_b": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.language_model.layers.45.self_attn.kv_a_proj_with_mqa": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.language_model.layers.45.self_attn.kv_b_proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.language_model.layers.45.self_attn.o_proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.language_model.layers.45.self_attn.q_a_proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.language_model.layers.45.self_attn.q_b_proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.language_model.layers.5.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.5.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.5.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.5.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.5.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.5.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.5.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.5.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.6.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.6.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.6.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.6.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.6.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.6.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.6.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.6.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.7.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.7.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.7.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.7.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.7.self_attn.kv_a_proj_with_mqa": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.7.self_attn.o_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.7.self_attn.q_a_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.7.self_attn.q_b_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.8.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.8.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.8.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.8.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.8.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.8.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.8.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.8.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.9.mlp.experts": {
+      "group_size": 16,
+      "quant_algo": "NVFP4",
+      "weight_scale_encoding": "csf_32x2_v1"
+    },
+    "model.language_model.layers.9.mlp.shared_experts.down_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.9.mlp.shared_experts.gate_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.9.mlp.shared_experts.up_proj": {
+      "quant_algo": "FP8_PB_WO",
+      "weight_block_size": [
+        128,
+        128
+      ]
+    },
+    "model.language_model.layers.9.self_attn.k_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.9.self_attn.o_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.9.self_attn.q_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.language_model.layers.9.self_attn.v_proj": {
+      "quant_algo": "MXFP8",
+      "weight_block_size": [
+        1,
+        32
+      ]
+    },
+    "model.visual.blocks.0.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.0.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.0.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.0.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.0.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.1.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.1.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.1.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.1.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.1.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.10.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.10.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.10.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.10.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.10.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.11.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.11.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.11.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.11.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.11.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.12.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.12.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.12.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.12.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.12.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.13.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.13.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.13.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.13.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.13.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.14.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.14.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.14.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.14.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.14.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.15.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.15.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.15.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.15.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.15.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.16.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.16.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.16.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.16.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.16.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.17.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.17.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.17.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.17.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.17.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.18.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.18.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.18.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.18.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.18.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.19.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.19.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.19.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.19.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.19.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.2.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.2.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.2.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.2.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.2.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.20.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.20.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.20.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.20.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.20.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.21.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.21.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.21.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.21.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.21.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.22.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.22.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.22.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.22.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.22.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.23.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.23.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.23.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.23.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.23.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.3.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.3.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.3.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.3.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.3.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.4.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.4.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.4.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.4.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.4.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.5.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.5.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.5.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.5.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.5.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.6.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.6.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.6.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.6.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.6.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.7.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.7.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.7.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.7.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.7.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.8.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.8.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.8.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.8.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.8.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.9.attn.proj": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.9.attn.qkv": {
+      "group_size": 32,
+      "quant_algo": "MXFP8"
+    },
+    "model.visual.blocks.9.mlp.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.9.mlp.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.blocks.9.mlp.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.merger.down_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.merger.gate_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.merger.proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    },
+    "model.visual.merger.up_proj": {
+      "group_size": 16,
+      "quant_algo": "W4A16_NVFP4"
+    }
+  }
+}
diff --git a/model-00001-of-00040.safetensors b/model-00001-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..43d5c2203aad3b208b0bd7a2f269afe5e006755d
--- /dev/null
+++ b/model-00001-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d2fcce2097ab08926290ff099b792c688a9eb014a6963ba201e6f6c18767730
+size 4294380516
diff --git a/model-00002-of-00040.safetensors b/model-00002-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..70926673ff0bb85c1395fe9e4a29284c865306e8
--- /dev/null
+++ b/model-00002-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abe5800514c86b4fe57a9f8419a1343460c5459495f626a15c143c077f0eed59
+size 398331164
diff --git a/model-00003-of-00040.safetensors b/model-00003-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..44c0535849673a487ee70ebad2239b6e6b2ac36f
--- /dev/null
+++ b/model-00003-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a6bf622464604faa5e7d8f3c7c265242fc768f90f5069bc5bff84d5665c3722f
+size 4187354632
diff --git a/model-00004-of-00040.safetensors b/model-00004-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..ef635d994b9df76e21673d4854e8eb7556380226
--- /dev/null
+++ b/model-00004-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1bf321c5b0c6a866f1916587cb84e0f67b6527fc9da1f432ce8ddb1eeefaf89f
+size 4059621056
diff --git a/model-00005-of-00040.safetensors b/model-00005-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..945bdd28a8764cde87c0ddcd4d100dda710a1fd6
--- /dev/null
+++ b/model-00005-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa93cdc2b83135bfd48c38a88e5d9299132e8b0d919ed2e68343fce8c7f1c844
+size 1204588600
diff --git a/model-00006-of-00040.safetensors b/model-00006-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..aac51400d70d8acf8f2a9d8181fcc69cecce8729
--- /dev/null
+++ b/model-00006-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f36a0aebce974e605a66c07c10963671753823640e8d54584a98fd04578414a
+size 3386693718
diff --git a/model-00007-of-00040.safetensors b/model-00007-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..9f871d4309c59b18a1563e210058f30f15398dfb
--- /dev/null
+++ b/model-00007-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08cf34e007f01060b73d63b9e73e3fb7429aef998e61bb521d927a5e1c836ca4
+size 3386630980
diff --git a/model-00008-of-00040.safetensors b/model-00008-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..3ef50edc9489d372bf0c9d3647c334906524280c
--- /dev/null
+++ b/model-00008-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27665caafba3db19034967c9f6c91e9614c0b71270a2a5983131ff845b34c66a
+size 3386784380
diff --git a/model-00009-of-00040.safetensors b/model-00009-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..ca62ef96ca94e23b438297aae53c79bdb3bb45a3
--- /dev/null
+++ b/model-00009-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fefef9818a88957a172129075db03351ff8705f912d187439650b548e882d1f
+size 3386714856
diff --git a/model-00010-of-00040.safetensors b/model-00010-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4ef01d298d9ac835b5a3d78f09395eff3ae40c83
--- /dev/null
+++ b/model-00010-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:618704500d946c3c573e9eca8ca74f0395139e190a76687feff6930201c32c02
+size 3386642180
diff --git a/model-00011-of-00040.safetensors b/model-00011-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e6672fdd95160b94ccd12aadad4ab5837d305ba6
--- /dev/null
+++ b/model-00011-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e79fd0c2161ffd83b02aba108713a0864959e26cb36cedaca5c1fecd1328d29b
+size 3386705796
diff --git a/model-00012-of-00040.safetensors b/model-00012-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d4fba74757d0a2b5ee6726cc9696b682733028dc
--- /dev/null
+++ b/model-00012-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc3ce32db330f4edaed1a814da5709e816b2b5874c582124cb2a319c123f0fb5
+size 3386704188
diff --git a/model-00013-of-00040.safetensors b/model-00013-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..0f7d0816bd58b21316e37988e9d2f88290619d54
--- /dev/null
+++ b/model-00013-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e44d1253f3fa790150f3c869c3fd04d692c08fcf29f54e9ed0f86f1e3804f44e
+size 3386634084
diff --git a/model-00014-of-00040.safetensors b/model-00014-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e72c45ebff16be8bf82f22288d95b86255823fd8
--- /dev/null
+++ b/model-00014-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9242f5c32541c58f9e9850091664f7359c6d4536254c5f808b68f133487404f
+size 3386785778
diff --git a/model-00015-of-00040.safetensors b/model-00015-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e879c8cc8e15ef1608a19af7178e5be4ae2ef5f9
--- /dev/null
+++ b/model-00015-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:856b041abe136f7bdf624df0e3f816999020e0ceaa44339f8f914cfe6423686c
+size 5080037446
diff --git a/model-00016-of-00040.safetensors b/model-00016-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..2109bdba05f59b15e986fea67748d028bc0db7a9
--- /dev/null
+++ b/model-00016-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4cc6b1eca1fd60ef05f4ed8b58fc31fc852fde068544fc1e06badcfae51e9dba
+size 5080062512
diff --git a/model-00017-of-00040.safetensors b/model-00017-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..0c2b92d138278aa437f11a9235f864205bc5dbe8
--- /dev/null
+++ b/model-00017-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2dc2ef526eccd2dc1df753aa6c22529a12340270b241f6429d915d43c8ffaf37
+size 5080033556
diff --git a/model-00018-of-00040.safetensors b/model-00018-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..9a35647d3f0475dfd7c4c72a724cd89591845361
--- /dev/null
+++ b/model-00018-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1fa1a86341b82892fb5ed234ad82d7836d3fcf004675a000e366116a423144c
+size 5080002106
diff --git a/model-00019-of-00040.safetensors b/model-00019-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e8bce0d5a789e56e5467bde3143fb81b662e26ca
--- /dev/null
+++ b/model-00019-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49ed0fd14feeb265c8c80e68b007feb6612f220b17cd7bec0382cf4fccdedad3
+size 5080044836
diff --git a/model-00020-of-00040.safetensors b/model-00020-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..0281e630d2a5f7e4c463b878e93aacabd2f4b0e7
--- /dev/null
+++ b/model-00020-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36d13ead5eff2bebc082ea8bfff2cd7952249e3d464e4691f5b33a7334095da8
+size 5080015448
diff --git a/model-00021-of-00040.safetensors b/model-00021-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..253fd14a1fe030f0ab280b80fd3606cc66686023
--- /dev/null
+++ b/model-00021-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be455e23d8170063b8ac2d22be604e71594fe72755c36b31272253cf8f94447c
+size 5080033720
diff --git a/model-00022-of-00040.safetensors b/model-00022-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..fd9b95b502b31c03b4b5dc872f5cfaa0bdb25d4f
--- /dev/null
+++ b/model-00022-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f7a0f6aeea5a2ff95b7c2f993675eca757cb3bd7e0ed9876850f73c0c640393
+size 5080007626
diff --git a/model-00023-of-00040.safetensors b/model-00023-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..de77bacdff3ef225e7ed3811f6feef1f02f31850
--- /dev/null
+++ b/model-00023-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18d7db72e3737476c604f6f23615318158e91026226eafcc3975d23293e36bbe
+size 5080036888
diff --git a/model-00024-of-00040.safetensors b/model-00024-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..8c3d1f08e571fac790ac7027a7693148db961598
--- /dev/null
+++ b/model-00024-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:937b0331965662e4d53e7266006bf9268dcf6e511354486396eb2e1cd1e1fce5
+size 5079998124
diff --git a/model-00025-of-00040.safetensors b/model-00025-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..24fb62b1961d506af075fd605d14a646bc75b858
--- /dev/null
+++ b/model-00025-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:522b327ece804e547f601a266be74cbdd4dfbc2938419052d43a5e1283ba94f5
+size 5080015080
diff --git a/model-00026-of-00040.safetensors b/model-00026-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..502f933c44f231648aeea5c81eb2491ffff9a4e7
--- /dev/null
+++ b/model-00026-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa5956b9758a382ff3a63a5c4625dd2a876183a8f6083bac1b665a7eb93903e1
+size 5080002452
diff --git a/model-00027-of-00040.safetensors b/model-00027-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..bfbed7bdeedf2f2f94cbc044355251d3de1a8d78
--- /dev/null
+++ b/model-00027-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e2b7685470b601aceebe226b26b52464fa40d74b9b346af337d45e8faacc1cc
+size 5080025960
diff --git a/model-00028-of-00040.safetensors b/model-00028-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..85df273bea48ccdecca65f0d7fc05c2fdb97260d
--- /dev/null
+++ b/model-00028-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83df1ae7d43eaf70b29fec3194851cb924775b8671440ed02ff2ac066cbaf7e5
+size 5080010694
diff --git a/model-00029-of-00040.safetensors b/model-00029-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..8eb1c04ed44b7404039c3c5e879707e7dcc1127b
--- /dev/null
+++ b/model-00029-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:223f763f1a1569d664f3e00641043bd6a0a0fe6c7881126620ef0183235fd5cd
+size 5080031820
diff --git a/model-00030-of-00040.safetensors b/model-00030-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..2047445ac782fc1caa6a7754549449f42c48fe69
--- /dev/null
+++ b/model-00030-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36e81695f6ddab8108aeb7479b7015a21b8ddb9057b4375527c13fb29af38392
+size 5080122474
diff --git a/model-00031-of-00040.safetensors b/model-00031-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..1f25b5aaa1c371c36f6039a5281003e72b34d131
--- /dev/null
+++ b/model-00031-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:697493ca2ce3f7fc975aaf9d8e65a2b373a2b4331035a4f5379ee4904688c73f
+size 5080041150
diff --git a/model-00032-of-00040.safetensors b/model-00032-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..ba5cc67f2bb28ae3a059ca4316349c950d636cab
--- /dev/null
+++ b/model-00032-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1967e72e60ed152b39ab0d951a3c524a885c6c2fb760e55688a08ccf197ae30b
+size 5080082340
diff --git a/model-00033-of-00040.safetensors b/model-00033-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..8eb19046cc088fed4ded4b913bba6658f7e005e7
--- /dev/null
+++ b/model-00033-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3896f4b763903405bfdf37249296bb05f525ca2134541dab7ea91c5513047a1c
+size 5080023370
diff --git a/model-00034-of-00040.safetensors b/model-00034-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..f6458f8148bed2d6c41acb9c0773a11920ffe4b2
--- /dev/null
+++ b/model-00034-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50d2cd8d608d3b8c59bd4c654275d6f840cd7598aeb42a2a5da8db75ae5c8b4d
+size 5079989500
diff --git a/model-00035-of-00040.safetensors b/model-00035-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d0f00ab455f4ad08511ed98ee17de2617376f125
--- /dev/null
+++ b/model-00035-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27ba1ee0f211a331fff5221ffe5f00b06eeb002f1865ae8d4fd8335d2c7f4c8c
+size 5080032548
diff --git a/model-00036-of-00040.safetensors b/model-00036-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..55b5205884912a44356adfbe343df585838b18b3
--- /dev/null
+++ b/model-00036-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6954778ce40d39efce3a07c01e1d245e757a701d72ead7a73e48fc2a57cbddd
+size 5080044714
diff --git a/model-00037-of-00040.safetensors b/model-00037-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..159c5306dd8d36ed5ab57c1b2c75ff8a1894986f
--- /dev/null
+++ b/model-00037-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a956ea3bd9468184565c5e94211c663bd401b3c1c57da08e96d5fd2b49419d9
+size 5080045746
diff --git a/model-00038-of-00040.safetensors b/model-00038-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..1e848e081e0087e8c78aed65a92cd383565339aa
--- /dev/null
+++ b/model-00038-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c12da94d3c1642f5ec34d1c1f573d193c75818bfc13b4774954cbd1ec9bfb646
+size 5080051898
diff --git a/model-00039-of-00040.safetensors b/model-00039-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..00dbff7fe29c2fe2c3b5c4b7f599341d09a63ab5
--- /dev/null
+++ b/model-00039-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5262d01f162e665c5da0f4e841d450e2396b4e45727b704c0ac1e514e287e7a9
+size 5080035462
diff --git a/model-00040-of-00040.safetensors b/model-00040-of-00040.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..5c0cffa925ba49042bada8a88dbc56b330bc7afb
--- /dev/null
+++ b/model-00040-of-00040.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73342fa305b49ec6677423b090479bbd8d021cd7e9f0afc27c2eeaa0b678dc23
+size 5080022666
diff --git a/model-inputscales.safetensors b/model-inputscales.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..fbbfd34d179e8ea1c303ed85f79ce37341fe88cb
--- /dev/null
+++ b/model-inputscales.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3dd6c79b081f1b61bd13a9354a0753f5b0406e3a4e8de05c45ff5c74e56ef1b
+size 4726696
diff --git a/model.safetensors.index.json b/model.safetensors.index.json
index edd2f873364e58d7b5f43df406b4d6abea769d2d..355f6df1caa8c92bf9cf528107f88ca56e70b5b2 100644
--- a/model.safetensors.index.json
+++ b/model.safetensors.index.json
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:568c770e4a083ab53c3d74deab348d541724f23ec9637d83513ecb8fa4a7ef73
-size 23449767
+oid sha256:1981941b2fabb9271ae3a3cb76be6ea41927f730b6633471d3c147106703aeea
+size 35266806
diff --git a/tensors/model-00001-of-00036.safetensors b/tensors/model-00001-of-00036.safetensors
deleted file mode 100644
index dd9427c1b63f66cde5a29572334a162ce48f0fed..0000000000000000000000000000000000000000
--- a/tensors/model-00001-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:bee2ce859f246fa647a28a6b8086d2d77c3388e878e135ec851c3c15f1283be4
-size 4997009796
diff --git a/tensors/model-00002-of-00036.safetensors b/tensors/model-00002-of-00036.safetensors
deleted file mode 100644
index e69189b09eb571728873678c56461b114f2f5924..0000000000000000000000000000000000000000
--- a/tensors/model-00002-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:58c17a9a8bb87bc108b2243d0b3554a0fd8e72a668f88bd4c050785a0716ada6
-size 4997194608
diff --git a/tensors/model-00003-of-00036.safetensors b/tensors/model-00003-of-00036.safetensors
deleted file mode 100644
index 37e1792904d731e7886e5d2a03869b0400874ee5..0000000000000000000000000000000000000000
--- a/tensors/model-00003-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5725f708d20f7089f589b020fa1f83f21d62e20c3c8ed5798aa2385c154b011d
-size 4997082968
diff --git a/tensors/model-00004-of-00036.safetensors b/tensors/model-00004-of-00036.safetensors
deleted file mode 100644
index e4035a11ee29aa67a3333456d8f968c2d460c9f3..0000000000000000000000000000000000000000
--- a/tensors/model-00004-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:15d33f9a63251c4fd63d9c7f9a2477003f1fcbbd5aaab4fc9f47261bf63def12
-size 4996991824
diff --git a/tensors/model-00005-of-00036.safetensors b/tensors/model-00005-of-00036.safetensors
deleted file mode 100644
index 0f161a0e67b7444c6b60ef5b19bd1d9b94d0dce1..0000000000000000000000000000000000000000
--- a/tensors/model-00005-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:8e7089fdbb56f76745f5ed1488d5610c181a77bab476b2c0cd18b83d52b650bf
-size 4997092464
diff --git a/tensors/model-00006-of-00036.safetensors b/tensors/model-00006-of-00036.safetensors
deleted file mode 100644
index d8ec3338ecb3f423148b3733a8e42012066058c3..0000000000000000000000000000000000000000
--- a/tensors/model-00006-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:5cc20bd94516fadc06153d81466401420cfc345b6bc193c948ce7c4bf020cec1
-size 4996967984
diff --git a/tensors/model-00007-of-00036.safetensors b/tensors/model-00007-of-00036.safetensors
deleted file mode 100644
index 327922207dc21a2eb91888680573828415f28567..0000000000000000000000000000000000000000
--- a/tensors/model-00007-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ac91e97bc6aa6cf55342bd91cfa03d20d0a61af7b7dcfb7eee157a9e51309372
-size 4997293804
diff --git a/tensors/model-00008-of-00036.safetensors b/tensors/model-00008-of-00036.safetensors
deleted file mode 100644
index e56f2c6998232ef134ce3aaf16cc5c105dbc2422..0000000000000000000000000000000000000000
--- a/tensors/model-00008-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:3fec2e3f9d1fdac465afb28d77e17bbfa721362aacca61e445c4d3db7fd64c43
-size 4996970628
diff --git a/tensors/model-00009-of-00036.safetensors b/tensors/model-00009-of-00036.safetensors
deleted file mode 100644
index 684f02fb8889e4ebf3b64e86c94dbfdf3ea152db..0000000000000000000000000000000000000000
--- a/tensors/model-00009-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:c075d2b39a2ba0e05d826fa35744bc453fa2550a3c8c23437c3b65bbc0c61082
-size 4997192808
diff --git a/tensors/model-00010-of-00036.safetensors b/tensors/model-00010-of-00036.safetensors
deleted file mode 100644
index 05c435356325b5c0580b0bc2b68d10e7faea7aeb..0000000000000000000000000000000000000000
--- a/tensors/model-00010-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:3e16cb82de5f953a6d88bf58de8f09037eada207faea50b0b4e5b97b4fe26e35
-size 4996948996
diff --git a/tensors/model-00011-of-00036.safetensors b/tensors/model-00011-of-00036.safetensors
deleted file mode 100644
index ff7b253ae45960eb6dd0def93714305edd961f0e..0000000000000000000000000000000000000000
--- a/tensors/model-00011-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:79913b9d9bb3c587b1152ef030f3140d0f7c8b1a089265e775bc0e2740ce90a6
-size 4997114488
diff --git a/tensors/model-00012-of-00036.safetensors b/tensors/model-00012-of-00036.safetensors
deleted file mode 100644
index f7565adb16c1120f6cdf5aa43b03ca22be9e68a0..0000000000000000000000000000000000000000
--- a/tensors/model-00012-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:399a88dc53644448c5ec8e8492c145618d51db0e7179e53fa7a88051567bad56
-size 4996944132
diff --git a/tensors/model-00013-of-00036.safetensors b/tensors/model-00013-of-00036.safetensors
deleted file mode 100644
index 025de8a64e15f0e50f5f4771fd90583c42731c30..0000000000000000000000000000000000000000
--- a/tensors/model-00013-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:557a3341555fc0e27ccd9446ea379358e5d74e97d7818bf8e9d2c10c832e4227
-size 4997096672
diff --git a/tensors/model-00014-of-00036.safetensors b/tensors/model-00014-of-00036.safetensors
deleted file mode 100644
index f2a5f0b3602af074f9e7e8fe96d75ae61f0980e5..0000000000000000000000000000000000000000
--- a/tensors/model-00014-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:37622fae7f9ee1aa538cdb0097ee13b971bf0e3ae620e81e1f4bcc86264d0a1c
-size 4996943364
diff --git a/tensors/model-00015-of-00036.safetensors b/tensors/model-00015-of-00036.safetensors
deleted file mode 100644
index fae48895cb1ad219fd7e1b61d9f61c10f99b11d6..0000000000000000000000000000000000000000
--- a/tensors/model-00015-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:f92f51d1df09d050a3166fee939e66bc607dccbf3e00fdbba3eebfaa38239333
-size 4997084344
diff --git a/tensors/model-00016-of-00036.safetensors b/tensors/model-00016-of-00036.safetensors
deleted file mode 100644
index 5cc4017cebc616183658363131cbd74f34b4306c..0000000000000000000000000000000000000000
--- a/tensors/model-00016-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:87e978c2abbcb39ae95e73ff366efbda5456639a81e098cd47b510db8f764a75
-size 4996913892
diff --git a/tensors/model-00017-of-00036.safetensors b/tensors/model-00017-of-00036.safetensors
deleted file mode 100644
index 161664bb93cc746aa8abdc285ef6786e68bc1e8e..0000000000000000000000000000000000000000
--- a/tensors/model-00017-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:096de82e005eb510096f72d3136dea6f80356728ac76ae808b69e2ca3c3f92a3
-size 4997065516
diff --git a/tensors/model-00018-of-00036.safetensors b/tensors/model-00018-of-00036.safetensors
deleted file mode 100644
index 9c8b02c9eca07488851c62eaa8cdd70969a3291f..0000000000000000000000000000000000000000
--- a/tensors/model-00018-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:d3c7d5d8e6a0f471d03800cf88c4d375945a687564b6f1f8bff7e519c830d564
-size 4996930276
diff --git a/tensors/model-00019-of-00036.safetensors b/tensors/model-00019-of-00036.safetensors
deleted file mode 100644
index 30d652dfb4b99cefc25ff4fe4423d085717781d9..0000000000000000000000000000000000000000
--- a/tensors/model-00019-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ff1cd2b20aff716f4314ed038ad0d356a3ccabb8c9eb983115a48859ce908ca1
-size 4997073008
diff --git a/tensors/model-00020-of-00036.safetensors b/tensors/model-00020-of-00036.safetensors
deleted file mode 100644
index 0a46268ac3f5b7a0184a38748d0ab523c75388db..0000000000000000000000000000000000000000
--- a/tensors/model-00020-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:d123dfc1206a1f6b64cf6398db5adc0584d353994fba7d257d0a9b18479444d8
-size 4996913640
diff --git a/tensors/model-00021-of-00036.safetensors b/tensors/model-00021-of-00036.safetensors
deleted file mode 100644
index 2cc9f35f32796aa302e9106a2f49495d647af82f..0000000000000000000000000000000000000000
--- a/tensors/model-00021-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:d391245f4fb5cdfd8d858da934f3360b658824679008ae9b423f0deb6d7e5e7c
-size 4997125272
diff --git a/tensors/model-00022-of-00036.safetensors b/tensors/model-00022-of-00036.safetensors
deleted file mode 100644
index 75a1ad6334f3a364a5ad2c5b31ae26c7b6d0d909..0000000000000000000000000000000000000000
--- a/tensors/model-00022-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:33bff3f9215815fd8131e492e4079585f2fdbd6979573cba219d00062f7e4678
-size 4996901328
diff --git a/tensors/model-00023-of-00036.safetensors b/tensors/model-00023-of-00036.safetensors
deleted file mode 100644
index d84203705da9cb34fed23461d3d2b59c41e2ad8c..0000000000000000000000000000000000000000
--- a/tensors/model-00023-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:75278b3533e644e6f203c5153fc668faaa6868d56a4581960e8115f7e818bedc
-size 4997355200
diff --git a/tensors/model-00024-of-00036.safetensors b/tensors/model-00024-of-00036.safetensors
deleted file mode 100644
index 0f666801a2f7c6b994cc5825c7ebee7951cbf01a..0000000000000000000000000000000000000000
--- a/tensors/model-00024-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ef6141efe7c20d241d43cf8821620901a43e5ef2f76b16846a2faf686af1299b
-size 4996951360
diff --git a/tensors/model-00025-of-00036.safetensors b/tensors/model-00025-of-00036.safetensors
deleted file mode 100644
index e86745ea876d802a039590c5e2779dcd7d741114..0000000000000000000000000000000000000000
--- a/tensors/model-00025-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:2f862d809c19c1e2e9cdda27d47d39006ea87a067dd64b34b6d1e25cd58a062e
-size 4997232564
diff --git a/tensors/model-00026-of-00036.safetensors b/tensors/model-00026-of-00036.safetensors
deleted file mode 100644
index 8773470cffa4bc282a17a9956c8d57a557cac98f..0000000000000000000000000000000000000000
--- a/tensors/model-00026-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:7c3ced571603a8d2027cb705bb49167628a798ef0dca2e6c2934fda4e20a95e9
-size 4996892836
diff --git a/tensors/model-00027-of-00036.safetensors b/tensors/model-00027-of-00036.safetensors
deleted file mode 100644
index a018f37593d519494037f0215fac398aba31db5a..0000000000000000000000000000000000000000
--- a/tensors/model-00027-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:cc059e4eae08dd20ba5f54e6becc399520e1428f4aa82a4c0866be5a769beabf
-size 4997073756
diff --git a/tensors/model-00028-of-00036.safetensors b/tensors/model-00028-of-00036.safetensors
deleted file mode 100644
index 1110c737850b9d2fbeafd67ffab3425d721066ad..0000000000000000000000000000000000000000
--- a/tensors/model-00028-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:cd5b23d3e9826239409bea852550cf0cffeba70257a29fdea53fc47c76ae3912
-size 4996930664
diff --git a/tensors/model-00029-of-00036.safetensors b/tensors/model-00029-of-00036.safetensors
deleted file mode 100644
index 673c0aded7d75948b78f30299cfc4fa6c49ad5cd..0000000000000000000000000000000000000000
--- a/tensors/model-00029-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:0c5d80cdba170ba39e63d0efa6052fcc03cf27918f59082a8440ba48d3040dbb
-size 4997188988
diff --git a/tensors/model-00030-of-00036.safetensors b/tensors/model-00030-of-00036.safetensors
deleted file mode 100644
index d291f90fe66a10b2511592719eeb81ff74070050..0000000000000000000000000000000000000000
--- a/tensors/model-00030-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:58156d30c3e67ca46d6d7549d0f979e32b2ae2d3cde3e799d9403f9de0c11031
-size 4996967996
diff --git a/tensors/model-00031-of-00036.safetensors b/tensors/model-00031-of-00036.safetensors
deleted file mode 100644
index a852c315597cb07588325a83fe599d28622f982b..0000000000000000000000000000000000000000
--- a/tensors/model-00031-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:2dc61b00fc16329751c58802b44d0cbfabce3c17a61a3630ca0f4123552df711
-size 4997170828
diff --git a/tensors/model-00032-of-00036.safetensors b/tensors/model-00032-of-00036.safetensors
deleted file mode 100644
index fc2d797b526079949aec1c86c460152b7301e9a1..0000000000000000000000000000000000000000
--- a/tensors/model-00032-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:4707ae80ba0b010891545e7eced8d0bec288ce7cf221b65cf9f5f141647d5c61
-size 4996950972
diff --git a/tensors/model-00033-of-00036.safetensors b/tensors/model-00033-of-00036.safetensors
deleted file mode 100644
index 04cf098a1809c7e27a06d6682987ae2bb6de4f00..0000000000000000000000000000000000000000
--- a/tensors/model-00033-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:856b635e1b073b5049b7700abd279690ba4dc624cf6822d4cc419a47f9fca1bd
-size 4988337164
diff --git a/tensors/model-00034-of-00036.safetensors b/tensors/model-00034-of-00036.safetensors
deleted file mode 100644
index 6cb42960173b220b22c5119dade43d79c9f07550..0000000000000000000000000000000000000000
--- a/tensors/model-00034-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:55d6b957a6fac947df848a844a520fe9305317ec8dae2ae4fa75c1601cd6ac73
-size 4999372640
diff --git a/tensors/model-00035-of-00036.safetensors b/tensors/model-00035-of-00036.safetensors
deleted file mode 100644
index 7d533d7aa4f3f3f71fb8841136e8e1a643ea8586..0000000000000000000000000000000000000000
--- a/tensors/model-00035-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:bbe43857d935891edc909248f10e3ce5eec43d50ea11d0a90425452b3975bd41
-size 5003932528
diff --git a/tensors/model-00036-of-00036.safetensors b/tensors/model-00036-of-00036.safetensors
deleted file mode 100644
index 006f03c24e3755c6af1561b72d9b1ff998d70e22..0000000000000000000000000000000000000000
--- a/tensors/model-00036-of-00036.safetensors
+++ /dev/null
@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:52caa105d4f466c774dbded16324a00c87cdb1e28782fc7dbb73cc9c10895770
-size 2809743712
diff --git a/tensors/native-QB.safetensors b/tensors/native-QB.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..c1549dec23b62de74b0f0ba63c121feec565308f
--- /dev/null
+++ b/tensors/native-QB.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3911ecc99b3a22a882b6cd892e7d948f099d069e61915dcfaa67b2cce00ffd41
+size 753692264