diff --git "a/config.json" "b/config.json" new file mode 100644--- /dev/null +++ "b/config.json" @@ -0,0 +1,7942 @@ +{ + "architectures": [ + "LongcatCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "attention_method": "MLA", + "bos_token_id": 1, + "cli_factor": 2, + "disable_quant_module": [], + "dsa_mtp_cli": true, + "eos_token_id": 2, + "expert_ffn_hidden_size": 2048, + "ffn_hidden_size": 12288, + "hidden_size": 8192, + "index_head_dim": 128, + "index_init_tokens": 16, + "index_k_norm_type": "rms", + "index_local_tokens": 1024, + "index_n_heads": 32, + "index_topk": 2048, + "kv_lora_rank": 512, + "max_position_embeddings": 262144, + "mla_scale_kv_lora": true, + "mla_scale_q_lora": true, + "model_type": "longcat2", + "moe_impl": "mix", + "moe_switch_token_num": 1024, + "moe_topk": 12, + "mtp_disable_over_tokenizer": true, + "mtp_num_layers": 3, + "mtp_replicate_modules": true, + "n_routed_experts": 768, + "num_attention_heads": 64, + "num_layers": 38, + "oe_neighbor_num": 5, + "oe_split_num": 4, + "oe_vocab_size_ratio": 100.567, + "q_lora_rank": 1536, + "qk_nope_head_dim": 128, + "qk_rope_head_dim": 64, + "quantization": { + "group_size": 64, + "bits": 3, + "mode": "affine", + "model.ngram_embeddings.word_embeddings": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.embedders.0": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.1": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.2": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.3": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.4": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.5": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.6": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.7": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.8": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.9": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.10": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.11": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.12": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.13": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.14": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.15": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.post_projs.0": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.1": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.2": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.3": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.4": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.5": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.6": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.7": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.8": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.9": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.10": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.11": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.12": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.13": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.14": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.15": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.0.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.0.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.0.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.0.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.0.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.0.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.0.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.1.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.1.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.1.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.1.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.1.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.1.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.1.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.2.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.2.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.2.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.2.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.2.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.2.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.2.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.3.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.3.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.3.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.3.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.3.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.3.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.3.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.4.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.4.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.4.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.4.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.4.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.4.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.4.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.5.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.5.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.5.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.5.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.5.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.5.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.5.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.6.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.6.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.6.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.6.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.6.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.6.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.6.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.7.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.7.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.7.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.7.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.7.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.7.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.7.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.8.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.8.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.8.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.8.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.8.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.8.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.8.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.9.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.9.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.9.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.9.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.9.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.9.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.9.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.10.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.10.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.10.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.10.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.10.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.10.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.10.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.11.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.11.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.11.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.11.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.11.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.11.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.11.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.12.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.12.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.12.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.12.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.12.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.12.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.12.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.13.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.13.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.13.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.13.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.13.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.13.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.13.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.14.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.14.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.14.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.14.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.14.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.14.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.14.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.15.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.15.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.15.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.15.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.15.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.15.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.15.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.16.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.16.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.16.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.16.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.16.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.16.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.16.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.17.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.17.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.17.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.17.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.17.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.17.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.17.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.18.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.18.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.18.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.18.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.18.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.18.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.18.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.19.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.19.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.19.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.19.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.19.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.19.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.19.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.20.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.20.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.20.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.20.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.20.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.20.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.20.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.21.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.21.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.21.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.21.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.21.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.21.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.21.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.22.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.22.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.22.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.22.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.22.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.22.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.22.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.23.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.23.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.23.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.23.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.23.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.23.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.23.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.24.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.24.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.24.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.24.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.24.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.24.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.24.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.25.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.25.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.25.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.25.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.25.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.25.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.25.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.26.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.26.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.26.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.26.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.26.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.26.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.26.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.27.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.27.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.27.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.27.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.27.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.27.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.27.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.28.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.28.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.28.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.28.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.28.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.28.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.28.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.29.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.29.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.29.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.29.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.29.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.29.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.29.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.30.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.30.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.30.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.30.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.30.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.30.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.30.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.31.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.31.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.31.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.31.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.31.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.31.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.31.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.32.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.32.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.32.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.32.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.32.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.32.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.32.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.33.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.33.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.33.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.33.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.33.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.33.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.33.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.34.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.34.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.34.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.34.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.34.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.34.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.34.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.35.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.35.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.35.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.35.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.35.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.35.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.35.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.36.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.36.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.36.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.36.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.36.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.36.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.36.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.37.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.37.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.37.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.37.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.37.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.37.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.37.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "lm_head": { + "group_size": 64, + "bits": 6 + } + }, + "quantization_config": { + "group_size": 64, + "bits": 3, + "mode": "affine", + "model.ngram_embeddings.word_embeddings": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.embedders.0": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.1": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.2": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.3": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.4": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.5": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.6": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.7": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.8": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.9": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.10": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.11": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.12": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.13": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.14": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.embedders.15": { + "group_size": 64, + "bits": 3 + }, + "model.ngram_embeddings.post_projs.0": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.1": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.2": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.3": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.4": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.5": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.6": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.7": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.8": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.9": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.10": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.11": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.12": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.13": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.14": { + "group_size": 64, + "bits": 6 + }, + "model.ngram_embeddings.post_projs.15": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.0.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.0.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.0.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.0.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.0.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.0.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.0.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.0.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.1.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.1.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.1.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.1.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.1.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.1.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.1.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.1.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.2.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.2.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.2.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.2.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.2.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.2.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.2.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.2.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.3.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.3.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.3.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.3.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.3.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.3.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.3.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.3.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.4.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.4.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.4.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.4.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.4.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.4.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.4.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.4.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.5.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.5.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.5.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.5.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.5.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.5.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.5.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.5.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.6.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.6.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.6.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.6.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.6.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.6.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.6.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.6.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.7.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.7.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.7.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.7.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.7.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.7.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.7.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.7.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.8.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.8.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.8.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.8.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.8.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.8.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.8.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.8.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.9.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.9.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.9.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.9.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.9.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.9.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.9.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.9.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.10.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.10.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.10.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.10.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.10.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.10.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.10.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.10.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.11.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.11.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.11.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.11.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.11.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.11.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.11.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.11.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.12.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.12.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.12.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.12.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.12.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.12.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.12.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.12.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.13.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.13.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.13.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.13.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.13.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.13.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.13.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.13.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.14.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.14.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.14.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.14.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.14.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.14.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.14.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.14.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.15.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.15.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.15.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.15.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.15.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.15.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.15.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.15.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.16.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.16.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.16.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.16.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.16.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.16.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.16.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.16.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.17.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.17.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.17.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.17.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.17.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.17.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.17.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.17.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.18.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.18.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.18.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.18.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.18.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.18.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.18.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.18.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.19.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.19.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.19.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.19.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.19.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.19.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.19.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.19.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.20.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.20.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.20.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.20.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.20.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.20.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.20.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.20.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.21.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.21.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.21.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.21.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.21.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.21.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.21.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.21.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.22.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.22.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.22.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.22.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.22.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.22.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.22.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.22.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.23.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.23.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.23.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.23.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.23.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.23.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.23.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.23.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.24.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.24.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.24.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.24.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.24.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.24.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.24.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.24.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.25.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.25.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.25.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.25.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.25.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.25.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.25.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.25.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.26.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.26.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.26.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.26.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.26.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.26.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.26.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.26.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.27.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.27.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.27.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.27.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.27.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.27.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.27.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.27.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.28.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.28.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.28.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.28.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.28.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.28.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.28.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.28.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.29.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.29.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.29.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.29.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.29.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.29.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.29.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.29.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.30.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.30.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.30.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.30.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.30.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.30.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.30.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.30.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.31.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.31.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.31.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.31.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.31.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.31.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.31.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.31.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.32.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.32.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.32.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.32.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.32.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.32.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.32.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.32.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.33.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.33.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.33.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.33.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.33.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.33.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.33.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.33.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.34.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.34.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.34.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.34.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.34.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.34.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.34.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.34.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.35.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.35.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.35.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.35.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.35.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.35.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.35.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.35.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.36.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.36.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.36.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.36.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.36.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.36.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.36.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.36.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlp.switch_mlp.gate_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.37.mlp.switch_mlp.up_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.37.mlp.switch_mlp.down_proj": { + "group_size": 64, + "bits": 3 + }, + "model.layers.37.mlp.router.classifier": { + "group_size": 64, + "bits": 8 + }, + "model.layers.37.self_attn.0.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.0.indexer.wq_b": { + "group_size": 64, + "bits": 8 + }, + "model.layers.37.self_attn.0.indexer.wk": { + "group_size": 64, + "bits": 8 + }, + "model.layers.37.self_attn.0.indexer.weights_proj": { + "group_size": 64, + "bits": 8 + }, + "model.layers.37.self_attn.1.q_a_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.1.q_b_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.1.kv_a_proj_with_mqa": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.1.embed_q": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.1.unembed_out": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.self_attn.1.o_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.0.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.0.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.0.down_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.1.gate_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.1.up_proj": { + "group_size": 64, + "bits": 6 + }, + "model.layers.37.mlps.1.down_proj": { + "group_size": 64, + "bits": 6 + }, + "lm_head": { + "group_size": 64, + "bits": 6 + } + }, + "rms_norm_eps": 1e-05, + "rope_scaling": { + "beta_fast": 32, + "beta_slow": 1, + "factor": 120, + "mscale": 1, + "mscale_all_dim": 1, + "original_max_position_embeddings": 8192, + "rope_type": "deepseek_yarn" + }, + "rope_theta": 1000000.0, + "routed_scaling_factor": 9, + "use_cache": true, + "use_mla": 1, + "v_head_dim": 128, + "vocab_size": 163840, + "zero_expert_num": 128, + "zero_expert_type": "identity" +} \ No newline at end of file