OLMo-3-7B-Think-VAC / config.json
Asystemoffields's picture
Upload folder using huggingface_hub
71b794c verified
Raw
History Blame Contribute Delete
39.1 kB
{
"transformers_version": "5.9.0",
"architectures": [
"Olmo3ForCausalLM"
],
"output_hidden_states": false,
"return_dict": true,
"dtype": "bfloat16",
"chunk_size_feed_forward": 0,
"is_encoder_decoder": false,
"id2label": {
"0": "LABEL_0",
"1": "LABEL_1"
},
"label2id": {
"LABEL_0": 0,
"LABEL_1": 1
},
"problem_type": null,
"vocab_size": 100278,
"hidden_size": 4096,
"intermediate_size": 11008,
"num_hidden_layers": 32,
"num_attention_heads": 32,
"num_key_value_heads": 32,
"hidden_act": "silu",
"max_position_embeddings": 65536,
"initializer_range": 0.02,
"use_cache": false,
"pad_token_id": 100277,
"bos_token_id": null,
"eos_token_id": 100257,
"tie_word_embeddings": false,
"rope_parameters": {
"attention_factor": 1.2079441541679836,
"beta_fast": 32.0,
"beta_slow": 1.0,
"factor": 8.0,
"original_max_position_embeddings": 8192,
"rope_type": "yarn",
"rope_theta": 500000
},
"attention_bias": false,
"attention_dropout": 0.0,
"rms_norm_eps": 1e-06,
"sliding_window": 4096,
"layer_types": [
"sliding_attention",
"sliding_attention",
"sliding_attention",
"full_attention",
"sliding_attention",
"sliding_attention",
"sliding_attention",
"full_attention",
"sliding_attention",
"sliding_attention",
"sliding_attention",
"full_attention",
"sliding_attention",
"sliding_attention",
"sliding_attention",
"full_attention",
"sliding_attention",
"sliding_attention",
"sliding_attention",
"full_attention",
"sliding_attention",
"sliding_attention",
"sliding_attention",
"full_attention",
"sliding_attention",
"sliding_attention",
"sliding_attention",
"full_attention",
"sliding_attention",
"sliding_attention",
"sliding_attention",
"full_attention"
],
"_name_or_path": "allenai/OLMo-3-7B-Think",
"model_type": "olmo3",
"output_attentions": false,
"auto_map": {
"AutoModelForCausalLM": "modeling_pmre_olmo.PmreOlmoForCausalLM"
},
"pmre_metadata": [
{
"module_path": "model.layers.0.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.0.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.0.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.0.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.0.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.0.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.0.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.1.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.1.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.1.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.1.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.1.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.1.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.1.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.2.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.2.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.2.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.2.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.2.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.2.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.2.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.3.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.3.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.3.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.3.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.3.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.3.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.3.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.4.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.4.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.4.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.4.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.4.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.4.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.4.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.5.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.5.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.5.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.5.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.5.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.5.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.5.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.6.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.6.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.6.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.6.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.6.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.6.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.6.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.7.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.7.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.7.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.7.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.7.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.7.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.7.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.8.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.8.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.8.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.8.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.8.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.8.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.8.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.9.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.9.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.9.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.9.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.9.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.9.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.9.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.10.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.10.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.10.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.10.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.10.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.10.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.10.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.11.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.11.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.11.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.11.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.11.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.11.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.11.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.12.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.12.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.12.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.12.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.12.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.12.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.12.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.13.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.13.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.13.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.13.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.13.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.13.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.13.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.14.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.14.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.14.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.14.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.14.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.14.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.14.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.15.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.15.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.15.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.15.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.15.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.15.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.15.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.16.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.16.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.16.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.16.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.16.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.16.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.16.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.17.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.17.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.17.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.17.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.17.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.17.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.17.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.18.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.18.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.18.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.18.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.18.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.18.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.18.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.19.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.19.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.19.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.19.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.19.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.19.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.19.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.20.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.20.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.20.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.20.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.20.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.20.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.20.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.21.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.21.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.21.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.21.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.21.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.21.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.21.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.22.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.22.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.22.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.22.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.22.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.22.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.22.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.23.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.23.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.23.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.23.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.23.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.23.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.23.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.24.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.24.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.24.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.24.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.24.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.24.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.24.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.25.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.25.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.25.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.25.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.25.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.25.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.25.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.26.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.26.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.26.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.26.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.26.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.26.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.26.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.27.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.27.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.27.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.27.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.27.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.27.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.27.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.28.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.28.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.28.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.28.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.28.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.28.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.28.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.29.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.29.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.29.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.29.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.29.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.29.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.29.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.30.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.30.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.30.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.30.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.30.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.30.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.30.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.31.self_attn.q_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.31.self_attn.k_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.31.self_attn.v_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.31.self_attn.o_proj",
"in_features": 4096,
"out_features": 4096,
"rank": 512,
"sparse_nnz": 0
},
{
"module_path": "model.layers.31.mlp.gate_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.31.mlp.up_proj",
"in_features": 4096,
"out_features": 11008,
"rank": 2132,
"sparse_nnz": 0
},
{
"module_path": "model.layers.31.mlp.down_proj",
"in_features": 11008,
"out_features": 4096,
"rank": 2132,
"sparse_nnz": 0
}
]
}