diff --git "a/config.json" "b/config.json" --- "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,