1476 lines
73 KiB
Plaintext
1476 lines
73 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 2,
|
|
"id": "5d13cd93",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"import torch\n",
|
|
"from sentence_transformers import SentenceTransformer\n",
|
|
"\n",
|
|
"\n",
|
|
"model_id = \"google/embeddinggemma-300M\"\n",
|
|
"model = SentenceTransformer(model_id)\n",
|
|
"\n",
|
|
"model_id = \"../../models/gemma/checkpoint-33246\"\n",
|
|
"model_lora = SentenceTransformer(model_id)\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 10,
|
|
"id": "ef587fd6",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"lora.Linear(\n",
|
|
" (base_layer): Linear(in_features=768, out_features=768, bias=False)\n",
|
|
" (lora_dropout): ModuleDict(\n",
|
|
" (default): Dropout(p=0.1, inplace=False)\n",
|
|
" )\n",
|
|
" (lora_A): ModuleDict(\n",
|
|
" (default): Linear(in_features=768, out_features=64, bias=False)\n",
|
|
" )\n",
|
|
" (lora_B): ModuleDict(\n",
|
|
" (default): Linear(in_features=64, out_features=768, bias=False)\n",
|
|
" )\n",
|
|
" (lora_embedding_A): ParameterDict()\n",
|
|
" (lora_embedding_B): ParameterDict()\n",
|
|
" (lora_magnitude_vector): ModuleDict()\n",
|
|
")"
|
|
]
|
|
},
|
|
"execution_count": 10,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"model_lora.get_submodule(\"0.auto_model.layers.0.self_attn.q_proj\")"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 8,
|
|
"id": "fd3e825d",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"Linear(in_features=768, out_features=768, bias=False)"
|
|
]
|
|
},
|
|
"execution_count": 8,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"model.get_submodule(\"0.auto_model.layers.0.self_attn.q_proj\")"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 3,
|
|
"id": "9fddb12a",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"\n",
|
|
"0\n",
|
|
"0.auto_model\n",
|
|
"0.auto_model.embed_tokens\n",
|
|
"0.auto_model.layers\n",
|
|
"0.auto_model.layers.0\n",
|
|
"0.auto_model.layers.0.self_attn\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.0.self_attn.k_proj\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.0.self_attn.o_proj\n",
|
|
"0.auto_model.layers.0.self_attn.q_norm\n",
|
|
"0.auto_model.layers.0.self_attn.k_norm\n",
|
|
"0.auto_model.layers.0.mlp\n",
|
|
"0.auto_model.layers.0.mlp.gate_proj\n",
|
|
"0.auto_model.layers.0.mlp.up_proj\n",
|
|
"0.auto_model.layers.0.mlp.down_proj\n",
|
|
"0.auto_model.layers.0.mlp.act_fn\n",
|
|
"0.auto_model.layers.0.input_layernorm\n",
|
|
"0.auto_model.layers.0.post_attention_layernorm\n",
|
|
"0.auto_model.layers.0.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.0.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.1\n",
|
|
"0.auto_model.layers.1.self_attn\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.1.self_attn.k_proj\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.1.self_attn.o_proj\n",
|
|
"0.auto_model.layers.1.self_attn.q_norm\n",
|
|
"0.auto_model.layers.1.self_attn.k_norm\n",
|
|
"0.auto_model.layers.1.mlp\n",
|
|
"0.auto_model.layers.1.mlp.gate_proj\n",
|
|
"0.auto_model.layers.1.mlp.up_proj\n",
|
|
"0.auto_model.layers.1.mlp.down_proj\n",
|
|
"0.auto_model.layers.1.mlp.act_fn\n",
|
|
"0.auto_model.layers.1.input_layernorm\n",
|
|
"0.auto_model.layers.1.post_attention_layernorm\n",
|
|
"0.auto_model.layers.1.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.1.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.2\n",
|
|
"0.auto_model.layers.2.self_attn\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.2.self_attn.k_proj\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.2.self_attn.o_proj\n",
|
|
"0.auto_model.layers.2.self_attn.q_norm\n",
|
|
"0.auto_model.layers.2.self_attn.k_norm\n",
|
|
"0.auto_model.layers.2.mlp\n",
|
|
"0.auto_model.layers.2.mlp.gate_proj\n",
|
|
"0.auto_model.layers.2.mlp.up_proj\n",
|
|
"0.auto_model.layers.2.mlp.down_proj\n",
|
|
"0.auto_model.layers.2.mlp.act_fn\n",
|
|
"0.auto_model.layers.2.input_layernorm\n",
|
|
"0.auto_model.layers.2.post_attention_layernorm\n",
|
|
"0.auto_model.layers.2.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.2.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.3\n",
|
|
"0.auto_model.layers.3.self_attn\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.3.self_attn.k_proj\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.3.self_attn.o_proj\n",
|
|
"0.auto_model.layers.3.self_attn.q_norm\n",
|
|
"0.auto_model.layers.3.self_attn.k_norm\n",
|
|
"0.auto_model.layers.3.mlp\n",
|
|
"0.auto_model.layers.3.mlp.gate_proj\n",
|
|
"0.auto_model.layers.3.mlp.up_proj\n",
|
|
"0.auto_model.layers.3.mlp.down_proj\n",
|
|
"0.auto_model.layers.3.mlp.act_fn\n",
|
|
"0.auto_model.layers.3.input_layernorm\n",
|
|
"0.auto_model.layers.3.post_attention_layernorm\n",
|
|
"0.auto_model.layers.3.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.3.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.4\n",
|
|
"0.auto_model.layers.4.self_attn\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.4.self_attn.k_proj\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.4.self_attn.o_proj\n",
|
|
"0.auto_model.layers.4.self_attn.q_norm\n",
|
|
"0.auto_model.layers.4.self_attn.k_norm\n",
|
|
"0.auto_model.layers.4.mlp\n",
|
|
"0.auto_model.layers.4.mlp.gate_proj\n",
|
|
"0.auto_model.layers.4.mlp.up_proj\n",
|
|
"0.auto_model.layers.4.mlp.down_proj\n",
|
|
"0.auto_model.layers.4.mlp.act_fn\n",
|
|
"0.auto_model.layers.4.input_layernorm\n",
|
|
"0.auto_model.layers.4.post_attention_layernorm\n",
|
|
"0.auto_model.layers.4.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.4.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.5\n",
|
|
"0.auto_model.layers.5.self_attn\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.5.self_attn.k_proj\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.5.self_attn.o_proj\n",
|
|
"0.auto_model.layers.5.self_attn.q_norm\n",
|
|
"0.auto_model.layers.5.self_attn.k_norm\n",
|
|
"0.auto_model.layers.5.mlp\n",
|
|
"0.auto_model.layers.5.mlp.gate_proj\n",
|
|
"0.auto_model.layers.5.mlp.up_proj\n",
|
|
"0.auto_model.layers.5.mlp.down_proj\n",
|
|
"0.auto_model.layers.5.mlp.act_fn\n",
|
|
"0.auto_model.layers.5.input_layernorm\n",
|
|
"0.auto_model.layers.5.post_attention_layernorm\n",
|
|
"0.auto_model.layers.5.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.5.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.6\n",
|
|
"0.auto_model.layers.6.self_attn\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.6.self_attn.k_proj\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.6.self_attn.o_proj\n",
|
|
"0.auto_model.layers.6.self_attn.q_norm\n",
|
|
"0.auto_model.layers.6.self_attn.k_norm\n",
|
|
"0.auto_model.layers.6.mlp\n",
|
|
"0.auto_model.layers.6.mlp.gate_proj\n",
|
|
"0.auto_model.layers.6.mlp.up_proj\n",
|
|
"0.auto_model.layers.6.mlp.down_proj\n",
|
|
"0.auto_model.layers.6.mlp.act_fn\n",
|
|
"0.auto_model.layers.6.input_layernorm\n",
|
|
"0.auto_model.layers.6.post_attention_layernorm\n",
|
|
"0.auto_model.layers.6.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.6.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.7\n",
|
|
"0.auto_model.layers.7.self_attn\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.7.self_attn.k_proj\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.7.self_attn.o_proj\n",
|
|
"0.auto_model.layers.7.self_attn.q_norm\n",
|
|
"0.auto_model.layers.7.self_attn.k_norm\n",
|
|
"0.auto_model.layers.7.mlp\n",
|
|
"0.auto_model.layers.7.mlp.gate_proj\n",
|
|
"0.auto_model.layers.7.mlp.up_proj\n",
|
|
"0.auto_model.layers.7.mlp.down_proj\n",
|
|
"0.auto_model.layers.7.mlp.act_fn\n",
|
|
"0.auto_model.layers.7.input_layernorm\n",
|
|
"0.auto_model.layers.7.post_attention_layernorm\n",
|
|
"0.auto_model.layers.7.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.7.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.8\n",
|
|
"0.auto_model.layers.8.self_attn\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.8.self_attn.k_proj\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.8.self_attn.o_proj\n",
|
|
"0.auto_model.layers.8.self_attn.q_norm\n",
|
|
"0.auto_model.layers.8.self_attn.k_norm\n",
|
|
"0.auto_model.layers.8.mlp\n",
|
|
"0.auto_model.layers.8.mlp.gate_proj\n",
|
|
"0.auto_model.layers.8.mlp.up_proj\n",
|
|
"0.auto_model.layers.8.mlp.down_proj\n",
|
|
"0.auto_model.layers.8.mlp.act_fn\n",
|
|
"0.auto_model.layers.8.input_layernorm\n",
|
|
"0.auto_model.layers.8.post_attention_layernorm\n",
|
|
"0.auto_model.layers.8.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.8.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.9\n",
|
|
"0.auto_model.layers.9.self_attn\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.9.self_attn.k_proj\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.9.self_attn.o_proj\n",
|
|
"0.auto_model.layers.9.self_attn.q_norm\n",
|
|
"0.auto_model.layers.9.self_attn.k_norm\n",
|
|
"0.auto_model.layers.9.mlp\n",
|
|
"0.auto_model.layers.9.mlp.gate_proj\n",
|
|
"0.auto_model.layers.9.mlp.up_proj\n",
|
|
"0.auto_model.layers.9.mlp.down_proj\n",
|
|
"0.auto_model.layers.9.mlp.act_fn\n",
|
|
"0.auto_model.layers.9.input_layernorm\n",
|
|
"0.auto_model.layers.9.post_attention_layernorm\n",
|
|
"0.auto_model.layers.9.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.9.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.10\n",
|
|
"0.auto_model.layers.10.self_attn\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.10.self_attn.k_proj\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.10.self_attn.o_proj\n",
|
|
"0.auto_model.layers.10.self_attn.q_norm\n",
|
|
"0.auto_model.layers.10.self_attn.k_norm\n",
|
|
"0.auto_model.layers.10.mlp\n",
|
|
"0.auto_model.layers.10.mlp.gate_proj\n",
|
|
"0.auto_model.layers.10.mlp.up_proj\n",
|
|
"0.auto_model.layers.10.mlp.down_proj\n",
|
|
"0.auto_model.layers.10.mlp.act_fn\n",
|
|
"0.auto_model.layers.10.input_layernorm\n",
|
|
"0.auto_model.layers.10.post_attention_layernorm\n",
|
|
"0.auto_model.layers.10.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.10.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.11\n",
|
|
"0.auto_model.layers.11.self_attn\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.11.self_attn.k_proj\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.11.self_attn.o_proj\n",
|
|
"0.auto_model.layers.11.self_attn.q_norm\n",
|
|
"0.auto_model.layers.11.self_attn.k_norm\n",
|
|
"0.auto_model.layers.11.mlp\n",
|
|
"0.auto_model.layers.11.mlp.gate_proj\n",
|
|
"0.auto_model.layers.11.mlp.up_proj\n",
|
|
"0.auto_model.layers.11.mlp.down_proj\n",
|
|
"0.auto_model.layers.11.mlp.act_fn\n",
|
|
"0.auto_model.layers.11.input_layernorm\n",
|
|
"0.auto_model.layers.11.post_attention_layernorm\n",
|
|
"0.auto_model.layers.11.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.11.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.12\n",
|
|
"0.auto_model.layers.12.self_attn\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.12.self_attn.k_proj\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.12.self_attn.o_proj\n",
|
|
"0.auto_model.layers.12.self_attn.q_norm\n",
|
|
"0.auto_model.layers.12.self_attn.k_norm\n",
|
|
"0.auto_model.layers.12.mlp\n",
|
|
"0.auto_model.layers.12.mlp.gate_proj\n",
|
|
"0.auto_model.layers.12.mlp.up_proj\n",
|
|
"0.auto_model.layers.12.mlp.down_proj\n",
|
|
"0.auto_model.layers.12.mlp.act_fn\n",
|
|
"0.auto_model.layers.12.input_layernorm\n",
|
|
"0.auto_model.layers.12.post_attention_layernorm\n",
|
|
"0.auto_model.layers.12.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.12.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.13\n",
|
|
"0.auto_model.layers.13.self_attn\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.13.self_attn.k_proj\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.13.self_attn.o_proj\n",
|
|
"0.auto_model.layers.13.self_attn.q_norm\n",
|
|
"0.auto_model.layers.13.self_attn.k_norm\n",
|
|
"0.auto_model.layers.13.mlp\n",
|
|
"0.auto_model.layers.13.mlp.gate_proj\n",
|
|
"0.auto_model.layers.13.mlp.up_proj\n",
|
|
"0.auto_model.layers.13.mlp.down_proj\n",
|
|
"0.auto_model.layers.13.mlp.act_fn\n",
|
|
"0.auto_model.layers.13.input_layernorm\n",
|
|
"0.auto_model.layers.13.post_attention_layernorm\n",
|
|
"0.auto_model.layers.13.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.13.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.14\n",
|
|
"0.auto_model.layers.14.self_attn\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.14.self_attn.k_proj\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.14.self_attn.o_proj\n",
|
|
"0.auto_model.layers.14.self_attn.q_norm\n",
|
|
"0.auto_model.layers.14.self_attn.k_norm\n",
|
|
"0.auto_model.layers.14.mlp\n",
|
|
"0.auto_model.layers.14.mlp.gate_proj\n",
|
|
"0.auto_model.layers.14.mlp.up_proj\n",
|
|
"0.auto_model.layers.14.mlp.down_proj\n",
|
|
"0.auto_model.layers.14.mlp.act_fn\n",
|
|
"0.auto_model.layers.14.input_layernorm\n",
|
|
"0.auto_model.layers.14.post_attention_layernorm\n",
|
|
"0.auto_model.layers.14.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.14.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.15\n",
|
|
"0.auto_model.layers.15.self_attn\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.15.self_attn.k_proj\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.15.self_attn.o_proj\n",
|
|
"0.auto_model.layers.15.self_attn.q_norm\n",
|
|
"0.auto_model.layers.15.self_attn.k_norm\n",
|
|
"0.auto_model.layers.15.mlp\n",
|
|
"0.auto_model.layers.15.mlp.gate_proj\n",
|
|
"0.auto_model.layers.15.mlp.up_proj\n",
|
|
"0.auto_model.layers.15.mlp.down_proj\n",
|
|
"0.auto_model.layers.15.mlp.act_fn\n",
|
|
"0.auto_model.layers.15.input_layernorm\n",
|
|
"0.auto_model.layers.15.post_attention_layernorm\n",
|
|
"0.auto_model.layers.15.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.15.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.16\n",
|
|
"0.auto_model.layers.16.self_attn\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.16.self_attn.k_proj\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.16.self_attn.o_proj\n",
|
|
"0.auto_model.layers.16.self_attn.q_norm\n",
|
|
"0.auto_model.layers.16.self_attn.k_norm\n",
|
|
"0.auto_model.layers.16.mlp\n",
|
|
"0.auto_model.layers.16.mlp.gate_proj\n",
|
|
"0.auto_model.layers.16.mlp.up_proj\n",
|
|
"0.auto_model.layers.16.mlp.down_proj\n",
|
|
"0.auto_model.layers.16.mlp.act_fn\n",
|
|
"0.auto_model.layers.16.input_layernorm\n",
|
|
"0.auto_model.layers.16.post_attention_layernorm\n",
|
|
"0.auto_model.layers.16.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.16.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.17\n",
|
|
"0.auto_model.layers.17.self_attn\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.17.self_attn.k_proj\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.17.self_attn.o_proj\n",
|
|
"0.auto_model.layers.17.self_attn.q_norm\n",
|
|
"0.auto_model.layers.17.self_attn.k_norm\n",
|
|
"0.auto_model.layers.17.mlp\n",
|
|
"0.auto_model.layers.17.mlp.gate_proj\n",
|
|
"0.auto_model.layers.17.mlp.up_proj\n",
|
|
"0.auto_model.layers.17.mlp.down_proj\n",
|
|
"0.auto_model.layers.17.mlp.act_fn\n",
|
|
"0.auto_model.layers.17.input_layernorm\n",
|
|
"0.auto_model.layers.17.post_attention_layernorm\n",
|
|
"0.auto_model.layers.17.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.17.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.18\n",
|
|
"0.auto_model.layers.18.self_attn\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.18.self_attn.k_proj\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.18.self_attn.o_proj\n",
|
|
"0.auto_model.layers.18.self_attn.q_norm\n",
|
|
"0.auto_model.layers.18.self_attn.k_norm\n",
|
|
"0.auto_model.layers.18.mlp\n",
|
|
"0.auto_model.layers.18.mlp.gate_proj\n",
|
|
"0.auto_model.layers.18.mlp.up_proj\n",
|
|
"0.auto_model.layers.18.mlp.down_proj\n",
|
|
"0.auto_model.layers.18.mlp.act_fn\n",
|
|
"0.auto_model.layers.18.input_layernorm\n",
|
|
"0.auto_model.layers.18.post_attention_layernorm\n",
|
|
"0.auto_model.layers.18.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.18.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.19\n",
|
|
"0.auto_model.layers.19.self_attn\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.19.self_attn.k_proj\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.19.self_attn.o_proj\n",
|
|
"0.auto_model.layers.19.self_attn.q_norm\n",
|
|
"0.auto_model.layers.19.self_attn.k_norm\n",
|
|
"0.auto_model.layers.19.mlp\n",
|
|
"0.auto_model.layers.19.mlp.gate_proj\n",
|
|
"0.auto_model.layers.19.mlp.up_proj\n",
|
|
"0.auto_model.layers.19.mlp.down_proj\n",
|
|
"0.auto_model.layers.19.mlp.act_fn\n",
|
|
"0.auto_model.layers.19.input_layernorm\n",
|
|
"0.auto_model.layers.19.post_attention_layernorm\n",
|
|
"0.auto_model.layers.19.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.19.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.20\n",
|
|
"0.auto_model.layers.20.self_attn\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.20.self_attn.k_proj\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.20.self_attn.o_proj\n",
|
|
"0.auto_model.layers.20.self_attn.q_norm\n",
|
|
"0.auto_model.layers.20.self_attn.k_norm\n",
|
|
"0.auto_model.layers.20.mlp\n",
|
|
"0.auto_model.layers.20.mlp.gate_proj\n",
|
|
"0.auto_model.layers.20.mlp.up_proj\n",
|
|
"0.auto_model.layers.20.mlp.down_proj\n",
|
|
"0.auto_model.layers.20.mlp.act_fn\n",
|
|
"0.auto_model.layers.20.input_layernorm\n",
|
|
"0.auto_model.layers.20.post_attention_layernorm\n",
|
|
"0.auto_model.layers.20.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.20.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.21\n",
|
|
"0.auto_model.layers.21.self_attn\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.21.self_attn.k_proj\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.21.self_attn.o_proj\n",
|
|
"0.auto_model.layers.21.self_attn.q_norm\n",
|
|
"0.auto_model.layers.21.self_attn.k_norm\n",
|
|
"0.auto_model.layers.21.mlp\n",
|
|
"0.auto_model.layers.21.mlp.gate_proj\n",
|
|
"0.auto_model.layers.21.mlp.up_proj\n",
|
|
"0.auto_model.layers.21.mlp.down_proj\n",
|
|
"0.auto_model.layers.21.mlp.act_fn\n",
|
|
"0.auto_model.layers.21.input_layernorm\n",
|
|
"0.auto_model.layers.21.post_attention_layernorm\n",
|
|
"0.auto_model.layers.21.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.21.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.22\n",
|
|
"0.auto_model.layers.22.self_attn\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.22.self_attn.k_proj\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.22.self_attn.o_proj\n",
|
|
"0.auto_model.layers.22.self_attn.q_norm\n",
|
|
"0.auto_model.layers.22.self_attn.k_norm\n",
|
|
"0.auto_model.layers.22.mlp\n",
|
|
"0.auto_model.layers.22.mlp.gate_proj\n",
|
|
"0.auto_model.layers.22.mlp.up_proj\n",
|
|
"0.auto_model.layers.22.mlp.down_proj\n",
|
|
"0.auto_model.layers.22.mlp.act_fn\n",
|
|
"0.auto_model.layers.22.input_layernorm\n",
|
|
"0.auto_model.layers.22.post_attention_layernorm\n",
|
|
"0.auto_model.layers.22.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.22.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.23\n",
|
|
"0.auto_model.layers.23.self_attn\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj.base_layer\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj.lora_dropout\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj.lora_A\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj.lora_A.default\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj.lora_B\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj.lora_B.default\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.23.self_attn.k_proj\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj.base_layer\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj.lora_dropout\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj.lora_dropout.default\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj.lora_A\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj.lora_A.default\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj.lora_B\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj.lora_B.default\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj.lora_embedding_A\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj.lora_embedding_B\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj.lora_magnitude_vector\n",
|
|
"0.auto_model.layers.23.self_attn.o_proj\n",
|
|
"0.auto_model.layers.23.self_attn.q_norm\n",
|
|
"0.auto_model.layers.23.self_attn.k_norm\n",
|
|
"0.auto_model.layers.23.mlp\n",
|
|
"0.auto_model.layers.23.mlp.gate_proj\n",
|
|
"0.auto_model.layers.23.mlp.up_proj\n",
|
|
"0.auto_model.layers.23.mlp.down_proj\n",
|
|
"0.auto_model.layers.23.mlp.act_fn\n",
|
|
"0.auto_model.layers.23.input_layernorm\n",
|
|
"0.auto_model.layers.23.post_attention_layernorm\n",
|
|
"0.auto_model.layers.23.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.23.post_feedforward_layernorm\n",
|
|
"0.auto_model.norm\n",
|
|
"0.auto_model.rotary_emb\n",
|
|
"0.auto_model.rotary_emb_local\n",
|
|
"1\n",
|
|
"2\n",
|
|
"2.activation_function\n",
|
|
"2.linear\n",
|
|
"3\n",
|
|
"3.activation_function\n",
|
|
"3.linear\n",
|
|
"4\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"for name, module in model_lora.named_modules():\n",
|
|
" print(name)\n",
|
|
"\n",
|
|
"\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 4,
|
|
"id": "891f30fb",
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"\n",
|
|
"0\n",
|
|
"0.auto_model\n",
|
|
"0.auto_model.embed_tokens\n",
|
|
"0.auto_model.layers\n",
|
|
"0.auto_model.layers.0\n",
|
|
"0.auto_model.layers.0.self_attn\n",
|
|
"0.auto_model.layers.0.self_attn.q_proj\n",
|
|
"0.auto_model.layers.0.self_attn.k_proj\n",
|
|
"0.auto_model.layers.0.self_attn.v_proj\n",
|
|
"0.auto_model.layers.0.self_attn.o_proj\n",
|
|
"0.auto_model.layers.0.self_attn.q_norm\n",
|
|
"0.auto_model.layers.0.self_attn.k_norm\n",
|
|
"0.auto_model.layers.0.mlp\n",
|
|
"0.auto_model.layers.0.mlp.gate_proj\n",
|
|
"0.auto_model.layers.0.mlp.up_proj\n",
|
|
"0.auto_model.layers.0.mlp.down_proj\n",
|
|
"0.auto_model.layers.0.mlp.act_fn\n",
|
|
"0.auto_model.layers.0.input_layernorm\n",
|
|
"0.auto_model.layers.0.post_attention_layernorm\n",
|
|
"0.auto_model.layers.0.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.0.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.1\n",
|
|
"0.auto_model.layers.1.self_attn\n",
|
|
"0.auto_model.layers.1.self_attn.q_proj\n",
|
|
"0.auto_model.layers.1.self_attn.k_proj\n",
|
|
"0.auto_model.layers.1.self_attn.v_proj\n",
|
|
"0.auto_model.layers.1.self_attn.o_proj\n",
|
|
"0.auto_model.layers.1.self_attn.q_norm\n",
|
|
"0.auto_model.layers.1.self_attn.k_norm\n",
|
|
"0.auto_model.layers.1.mlp\n",
|
|
"0.auto_model.layers.1.mlp.gate_proj\n",
|
|
"0.auto_model.layers.1.mlp.up_proj\n",
|
|
"0.auto_model.layers.1.mlp.down_proj\n",
|
|
"0.auto_model.layers.1.mlp.act_fn\n",
|
|
"0.auto_model.layers.1.input_layernorm\n",
|
|
"0.auto_model.layers.1.post_attention_layernorm\n",
|
|
"0.auto_model.layers.1.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.1.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.2\n",
|
|
"0.auto_model.layers.2.self_attn\n",
|
|
"0.auto_model.layers.2.self_attn.q_proj\n",
|
|
"0.auto_model.layers.2.self_attn.k_proj\n",
|
|
"0.auto_model.layers.2.self_attn.v_proj\n",
|
|
"0.auto_model.layers.2.self_attn.o_proj\n",
|
|
"0.auto_model.layers.2.self_attn.q_norm\n",
|
|
"0.auto_model.layers.2.self_attn.k_norm\n",
|
|
"0.auto_model.layers.2.mlp\n",
|
|
"0.auto_model.layers.2.mlp.gate_proj\n",
|
|
"0.auto_model.layers.2.mlp.up_proj\n",
|
|
"0.auto_model.layers.2.mlp.down_proj\n",
|
|
"0.auto_model.layers.2.mlp.act_fn\n",
|
|
"0.auto_model.layers.2.input_layernorm\n",
|
|
"0.auto_model.layers.2.post_attention_layernorm\n",
|
|
"0.auto_model.layers.2.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.2.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.3\n",
|
|
"0.auto_model.layers.3.self_attn\n",
|
|
"0.auto_model.layers.3.self_attn.q_proj\n",
|
|
"0.auto_model.layers.3.self_attn.k_proj\n",
|
|
"0.auto_model.layers.3.self_attn.v_proj\n",
|
|
"0.auto_model.layers.3.self_attn.o_proj\n",
|
|
"0.auto_model.layers.3.self_attn.q_norm\n",
|
|
"0.auto_model.layers.3.self_attn.k_norm\n",
|
|
"0.auto_model.layers.3.mlp\n",
|
|
"0.auto_model.layers.3.mlp.gate_proj\n",
|
|
"0.auto_model.layers.3.mlp.up_proj\n",
|
|
"0.auto_model.layers.3.mlp.down_proj\n",
|
|
"0.auto_model.layers.3.mlp.act_fn\n",
|
|
"0.auto_model.layers.3.input_layernorm\n",
|
|
"0.auto_model.layers.3.post_attention_layernorm\n",
|
|
"0.auto_model.layers.3.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.3.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.4\n",
|
|
"0.auto_model.layers.4.self_attn\n",
|
|
"0.auto_model.layers.4.self_attn.q_proj\n",
|
|
"0.auto_model.layers.4.self_attn.k_proj\n",
|
|
"0.auto_model.layers.4.self_attn.v_proj\n",
|
|
"0.auto_model.layers.4.self_attn.o_proj\n",
|
|
"0.auto_model.layers.4.self_attn.q_norm\n",
|
|
"0.auto_model.layers.4.self_attn.k_norm\n",
|
|
"0.auto_model.layers.4.mlp\n",
|
|
"0.auto_model.layers.4.mlp.gate_proj\n",
|
|
"0.auto_model.layers.4.mlp.up_proj\n",
|
|
"0.auto_model.layers.4.mlp.down_proj\n",
|
|
"0.auto_model.layers.4.mlp.act_fn\n",
|
|
"0.auto_model.layers.4.input_layernorm\n",
|
|
"0.auto_model.layers.4.post_attention_layernorm\n",
|
|
"0.auto_model.layers.4.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.4.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.5\n",
|
|
"0.auto_model.layers.5.self_attn\n",
|
|
"0.auto_model.layers.5.self_attn.q_proj\n",
|
|
"0.auto_model.layers.5.self_attn.k_proj\n",
|
|
"0.auto_model.layers.5.self_attn.v_proj\n",
|
|
"0.auto_model.layers.5.self_attn.o_proj\n",
|
|
"0.auto_model.layers.5.self_attn.q_norm\n",
|
|
"0.auto_model.layers.5.self_attn.k_norm\n",
|
|
"0.auto_model.layers.5.mlp\n",
|
|
"0.auto_model.layers.5.mlp.gate_proj\n",
|
|
"0.auto_model.layers.5.mlp.up_proj\n",
|
|
"0.auto_model.layers.5.mlp.down_proj\n",
|
|
"0.auto_model.layers.5.mlp.act_fn\n",
|
|
"0.auto_model.layers.5.input_layernorm\n",
|
|
"0.auto_model.layers.5.post_attention_layernorm\n",
|
|
"0.auto_model.layers.5.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.5.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.6\n",
|
|
"0.auto_model.layers.6.self_attn\n",
|
|
"0.auto_model.layers.6.self_attn.q_proj\n",
|
|
"0.auto_model.layers.6.self_attn.k_proj\n",
|
|
"0.auto_model.layers.6.self_attn.v_proj\n",
|
|
"0.auto_model.layers.6.self_attn.o_proj\n",
|
|
"0.auto_model.layers.6.self_attn.q_norm\n",
|
|
"0.auto_model.layers.6.self_attn.k_norm\n",
|
|
"0.auto_model.layers.6.mlp\n",
|
|
"0.auto_model.layers.6.mlp.gate_proj\n",
|
|
"0.auto_model.layers.6.mlp.up_proj\n",
|
|
"0.auto_model.layers.6.mlp.down_proj\n",
|
|
"0.auto_model.layers.6.mlp.act_fn\n",
|
|
"0.auto_model.layers.6.input_layernorm\n",
|
|
"0.auto_model.layers.6.post_attention_layernorm\n",
|
|
"0.auto_model.layers.6.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.6.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.7\n",
|
|
"0.auto_model.layers.7.self_attn\n",
|
|
"0.auto_model.layers.7.self_attn.q_proj\n",
|
|
"0.auto_model.layers.7.self_attn.k_proj\n",
|
|
"0.auto_model.layers.7.self_attn.v_proj\n",
|
|
"0.auto_model.layers.7.self_attn.o_proj\n",
|
|
"0.auto_model.layers.7.self_attn.q_norm\n",
|
|
"0.auto_model.layers.7.self_attn.k_norm\n",
|
|
"0.auto_model.layers.7.mlp\n",
|
|
"0.auto_model.layers.7.mlp.gate_proj\n",
|
|
"0.auto_model.layers.7.mlp.up_proj\n",
|
|
"0.auto_model.layers.7.mlp.down_proj\n",
|
|
"0.auto_model.layers.7.mlp.act_fn\n",
|
|
"0.auto_model.layers.7.input_layernorm\n",
|
|
"0.auto_model.layers.7.post_attention_layernorm\n",
|
|
"0.auto_model.layers.7.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.7.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.8\n",
|
|
"0.auto_model.layers.8.self_attn\n",
|
|
"0.auto_model.layers.8.self_attn.q_proj\n",
|
|
"0.auto_model.layers.8.self_attn.k_proj\n",
|
|
"0.auto_model.layers.8.self_attn.v_proj\n",
|
|
"0.auto_model.layers.8.self_attn.o_proj\n",
|
|
"0.auto_model.layers.8.self_attn.q_norm\n",
|
|
"0.auto_model.layers.8.self_attn.k_norm\n",
|
|
"0.auto_model.layers.8.mlp\n",
|
|
"0.auto_model.layers.8.mlp.gate_proj\n",
|
|
"0.auto_model.layers.8.mlp.up_proj\n",
|
|
"0.auto_model.layers.8.mlp.down_proj\n",
|
|
"0.auto_model.layers.8.mlp.act_fn\n",
|
|
"0.auto_model.layers.8.input_layernorm\n",
|
|
"0.auto_model.layers.8.post_attention_layernorm\n",
|
|
"0.auto_model.layers.8.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.8.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.9\n",
|
|
"0.auto_model.layers.9.self_attn\n",
|
|
"0.auto_model.layers.9.self_attn.q_proj\n",
|
|
"0.auto_model.layers.9.self_attn.k_proj\n",
|
|
"0.auto_model.layers.9.self_attn.v_proj\n",
|
|
"0.auto_model.layers.9.self_attn.o_proj\n",
|
|
"0.auto_model.layers.9.self_attn.q_norm\n",
|
|
"0.auto_model.layers.9.self_attn.k_norm\n",
|
|
"0.auto_model.layers.9.mlp\n",
|
|
"0.auto_model.layers.9.mlp.gate_proj\n",
|
|
"0.auto_model.layers.9.mlp.up_proj\n",
|
|
"0.auto_model.layers.9.mlp.down_proj\n",
|
|
"0.auto_model.layers.9.mlp.act_fn\n",
|
|
"0.auto_model.layers.9.input_layernorm\n",
|
|
"0.auto_model.layers.9.post_attention_layernorm\n",
|
|
"0.auto_model.layers.9.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.9.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.10\n",
|
|
"0.auto_model.layers.10.self_attn\n",
|
|
"0.auto_model.layers.10.self_attn.q_proj\n",
|
|
"0.auto_model.layers.10.self_attn.k_proj\n",
|
|
"0.auto_model.layers.10.self_attn.v_proj\n",
|
|
"0.auto_model.layers.10.self_attn.o_proj\n",
|
|
"0.auto_model.layers.10.self_attn.q_norm\n",
|
|
"0.auto_model.layers.10.self_attn.k_norm\n",
|
|
"0.auto_model.layers.10.mlp\n",
|
|
"0.auto_model.layers.10.mlp.gate_proj\n",
|
|
"0.auto_model.layers.10.mlp.up_proj\n",
|
|
"0.auto_model.layers.10.mlp.down_proj\n",
|
|
"0.auto_model.layers.10.mlp.act_fn\n",
|
|
"0.auto_model.layers.10.input_layernorm\n",
|
|
"0.auto_model.layers.10.post_attention_layernorm\n",
|
|
"0.auto_model.layers.10.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.10.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.11\n",
|
|
"0.auto_model.layers.11.self_attn\n",
|
|
"0.auto_model.layers.11.self_attn.q_proj\n",
|
|
"0.auto_model.layers.11.self_attn.k_proj\n",
|
|
"0.auto_model.layers.11.self_attn.v_proj\n",
|
|
"0.auto_model.layers.11.self_attn.o_proj\n",
|
|
"0.auto_model.layers.11.self_attn.q_norm\n",
|
|
"0.auto_model.layers.11.self_attn.k_norm\n",
|
|
"0.auto_model.layers.11.mlp\n",
|
|
"0.auto_model.layers.11.mlp.gate_proj\n",
|
|
"0.auto_model.layers.11.mlp.up_proj\n",
|
|
"0.auto_model.layers.11.mlp.down_proj\n",
|
|
"0.auto_model.layers.11.mlp.act_fn\n",
|
|
"0.auto_model.layers.11.input_layernorm\n",
|
|
"0.auto_model.layers.11.post_attention_layernorm\n",
|
|
"0.auto_model.layers.11.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.11.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.12\n",
|
|
"0.auto_model.layers.12.self_attn\n",
|
|
"0.auto_model.layers.12.self_attn.q_proj\n",
|
|
"0.auto_model.layers.12.self_attn.k_proj\n",
|
|
"0.auto_model.layers.12.self_attn.v_proj\n",
|
|
"0.auto_model.layers.12.self_attn.o_proj\n",
|
|
"0.auto_model.layers.12.self_attn.q_norm\n",
|
|
"0.auto_model.layers.12.self_attn.k_norm\n",
|
|
"0.auto_model.layers.12.mlp\n",
|
|
"0.auto_model.layers.12.mlp.gate_proj\n",
|
|
"0.auto_model.layers.12.mlp.up_proj\n",
|
|
"0.auto_model.layers.12.mlp.down_proj\n",
|
|
"0.auto_model.layers.12.mlp.act_fn\n",
|
|
"0.auto_model.layers.12.input_layernorm\n",
|
|
"0.auto_model.layers.12.post_attention_layernorm\n",
|
|
"0.auto_model.layers.12.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.12.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.13\n",
|
|
"0.auto_model.layers.13.self_attn\n",
|
|
"0.auto_model.layers.13.self_attn.q_proj\n",
|
|
"0.auto_model.layers.13.self_attn.k_proj\n",
|
|
"0.auto_model.layers.13.self_attn.v_proj\n",
|
|
"0.auto_model.layers.13.self_attn.o_proj\n",
|
|
"0.auto_model.layers.13.self_attn.q_norm\n",
|
|
"0.auto_model.layers.13.self_attn.k_norm\n",
|
|
"0.auto_model.layers.13.mlp\n",
|
|
"0.auto_model.layers.13.mlp.gate_proj\n",
|
|
"0.auto_model.layers.13.mlp.up_proj\n",
|
|
"0.auto_model.layers.13.mlp.down_proj\n",
|
|
"0.auto_model.layers.13.mlp.act_fn\n",
|
|
"0.auto_model.layers.13.input_layernorm\n",
|
|
"0.auto_model.layers.13.post_attention_layernorm\n",
|
|
"0.auto_model.layers.13.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.13.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.14\n",
|
|
"0.auto_model.layers.14.self_attn\n",
|
|
"0.auto_model.layers.14.self_attn.q_proj\n",
|
|
"0.auto_model.layers.14.self_attn.k_proj\n",
|
|
"0.auto_model.layers.14.self_attn.v_proj\n",
|
|
"0.auto_model.layers.14.self_attn.o_proj\n",
|
|
"0.auto_model.layers.14.self_attn.q_norm\n",
|
|
"0.auto_model.layers.14.self_attn.k_norm\n",
|
|
"0.auto_model.layers.14.mlp\n",
|
|
"0.auto_model.layers.14.mlp.gate_proj\n",
|
|
"0.auto_model.layers.14.mlp.up_proj\n",
|
|
"0.auto_model.layers.14.mlp.down_proj\n",
|
|
"0.auto_model.layers.14.mlp.act_fn\n",
|
|
"0.auto_model.layers.14.input_layernorm\n",
|
|
"0.auto_model.layers.14.post_attention_layernorm\n",
|
|
"0.auto_model.layers.14.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.14.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.15\n",
|
|
"0.auto_model.layers.15.self_attn\n",
|
|
"0.auto_model.layers.15.self_attn.q_proj\n",
|
|
"0.auto_model.layers.15.self_attn.k_proj\n",
|
|
"0.auto_model.layers.15.self_attn.v_proj\n",
|
|
"0.auto_model.layers.15.self_attn.o_proj\n",
|
|
"0.auto_model.layers.15.self_attn.q_norm\n",
|
|
"0.auto_model.layers.15.self_attn.k_norm\n",
|
|
"0.auto_model.layers.15.mlp\n",
|
|
"0.auto_model.layers.15.mlp.gate_proj\n",
|
|
"0.auto_model.layers.15.mlp.up_proj\n",
|
|
"0.auto_model.layers.15.mlp.down_proj\n",
|
|
"0.auto_model.layers.15.mlp.act_fn\n",
|
|
"0.auto_model.layers.15.input_layernorm\n",
|
|
"0.auto_model.layers.15.post_attention_layernorm\n",
|
|
"0.auto_model.layers.15.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.15.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.16\n",
|
|
"0.auto_model.layers.16.self_attn\n",
|
|
"0.auto_model.layers.16.self_attn.q_proj\n",
|
|
"0.auto_model.layers.16.self_attn.k_proj\n",
|
|
"0.auto_model.layers.16.self_attn.v_proj\n",
|
|
"0.auto_model.layers.16.self_attn.o_proj\n",
|
|
"0.auto_model.layers.16.self_attn.q_norm\n",
|
|
"0.auto_model.layers.16.self_attn.k_norm\n",
|
|
"0.auto_model.layers.16.mlp\n",
|
|
"0.auto_model.layers.16.mlp.gate_proj\n",
|
|
"0.auto_model.layers.16.mlp.up_proj\n",
|
|
"0.auto_model.layers.16.mlp.down_proj\n",
|
|
"0.auto_model.layers.16.mlp.act_fn\n",
|
|
"0.auto_model.layers.16.input_layernorm\n",
|
|
"0.auto_model.layers.16.post_attention_layernorm\n",
|
|
"0.auto_model.layers.16.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.16.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.17\n",
|
|
"0.auto_model.layers.17.self_attn\n",
|
|
"0.auto_model.layers.17.self_attn.q_proj\n",
|
|
"0.auto_model.layers.17.self_attn.k_proj\n",
|
|
"0.auto_model.layers.17.self_attn.v_proj\n",
|
|
"0.auto_model.layers.17.self_attn.o_proj\n",
|
|
"0.auto_model.layers.17.self_attn.q_norm\n",
|
|
"0.auto_model.layers.17.self_attn.k_norm\n",
|
|
"0.auto_model.layers.17.mlp\n",
|
|
"0.auto_model.layers.17.mlp.gate_proj\n",
|
|
"0.auto_model.layers.17.mlp.up_proj\n",
|
|
"0.auto_model.layers.17.mlp.down_proj\n",
|
|
"0.auto_model.layers.17.mlp.act_fn\n",
|
|
"0.auto_model.layers.17.input_layernorm\n",
|
|
"0.auto_model.layers.17.post_attention_layernorm\n",
|
|
"0.auto_model.layers.17.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.17.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.18\n",
|
|
"0.auto_model.layers.18.self_attn\n",
|
|
"0.auto_model.layers.18.self_attn.q_proj\n",
|
|
"0.auto_model.layers.18.self_attn.k_proj\n",
|
|
"0.auto_model.layers.18.self_attn.v_proj\n",
|
|
"0.auto_model.layers.18.self_attn.o_proj\n",
|
|
"0.auto_model.layers.18.self_attn.q_norm\n",
|
|
"0.auto_model.layers.18.self_attn.k_norm\n",
|
|
"0.auto_model.layers.18.mlp\n",
|
|
"0.auto_model.layers.18.mlp.gate_proj\n",
|
|
"0.auto_model.layers.18.mlp.up_proj\n",
|
|
"0.auto_model.layers.18.mlp.down_proj\n",
|
|
"0.auto_model.layers.18.mlp.act_fn\n",
|
|
"0.auto_model.layers.18.input_layernorm\n",
|
|
"0.auto_model.layers.18.post_attention_layernorm\n",
|
|
"0.auto_model.layers.18.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.18.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.19\n",
|
|
"0.auto_model.layers.19.self_attn\n",
|
|
"0.auto_model.layers.19.self_attn.q_proj\n",
|
|
"0.auto_model.layers.19.self_attn.k_proj\n",
|
|
"0.auto_model.layers.19.self_attn.v_proj\n",
|
|
"0.auto_model.layers.19.self_attn.o_proj\n",
|
|
"0.auto_model.layers.19.self_attn.q_norm\n",
|
|
"0.auto_model.layers.19.self_attn.k_norm\n",
|
|
"0.auto_model.layers.19.mlp\n",
|
|
"0.auto_model.layers.19.mlp.gate_proj\n",
|
|
"0.auto_model.layers.19.mlp.up_proj\n",
|
|
"0.auto_model.layers.19.mlp.down_proj\n",
|
|
"0.auto_model.layers.19.mlp.act_fn\n",
|
|
"0.auto_model.layers.19.input_layernorm\n",
|
|
"0.auto_model.layers.19.post_attention_layernorm\n",
|
|
"0.auto_model.layers.19.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.19.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.20\n",
|
|
"0.auto_model.layers.20.self_attn\n",
|
|
"0.auto_model.layers.20.self_attn.q_proj\n",
|
|
"0.auto_model.layers.20.self_attn.k_proj\n",
|
|
"0.auto_model.layers.20.self_attn.v_proj\n",
|
|
"0.auto_model.layers.20.self_attn.o_proj\n",
|
|
"0.auto_model.layers.20.self_attn.q_norm\n",
|
|
"0.auto_model.layers.20.self_attn.k_norm\n",
|
|
"0.auto_model.layers.20.mlp\n",
|
|
"0.auto_model.layers.20.mlp.gate_proj\n",
|
|
"0.auto_model.layers.20.mlp.up_proj\n",
|
|
"0.auto_model.layers.20.mlp.down_proj\n",
|
|
"0.auto_model.layers.20.mlp.act_fn\n",
|
|
"0.auto_model.layers.20.input_layernorm\n",
|
|
"0.auto_model.layers.20.post_attention_layernorm\n",
|
|
"0.auto_model.layers.20.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.20.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.21\n",
|
|
"0.auto_model.layers.21.self_attn\n",
|
|
"0.auto_model.layers.21.self_attn.q_proj\n",
|
|
"0.auto_model.layers.21.self_attn.k_proj\n",
|
|
"0.auto_model.layers.21.self_attn.v_proj\n",
|
|
"0.auto_model.layers.21.self_attn.o_proj\n",
|
|
"0.auto_model.layers.21.self_attn.q_norm\n",
|
|
"0.auto_model.layers.21.self_attn.k_norm\n",
|
|
"0.auto_model.layers.21.mlp\n",
|
|
"0.auto_model.layers.21.mlp.gate_proj\n",
|
|
"0.auto_model.layers.21.mlp.up_proj\n",
|
|
"0.auto_model.layers.21.mlp.down_proj\n",
|
|
"0.auto_model.layers.21.mlp.act_fn\n",
|
|
"0.auto_model.layers.21.input_layernorm\n",
|
|
"0.auto_model.layers.21.post_attention_layernorm\n",
|
|
"0.auto_model.layers.21.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.21.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.22\n",
|
|
"0.auto_model.layers.22.self_attn\n",
|
|
"0.auto_model.layers.22.self_attn.q_proj\n",
|
|
"0.auto_model.layers.22.self_attn.k_proj\n",
|
|
"0.auto_model.layers.22.self_attn.v_proj\n",
|
|
"0.auto_model.layers.22.self_attn.o_proj\n",
|
|
"0.auto_model.layers.22.self_attn.q_norm\n",
|
|
"0.auto_model.layers.22.self_attn.k_norm\n",
|
|
"0.auto_model.layers.22.mlp\n",
|
|
"0.auto_model.layers.22.mlp.gate_proj\n",
|
|
"0.auto_model.layers.22.mlp.up_proj\n",
|
|
"0.auto_model.layers.22.mlp.down_proj\n",
|
|
"0.auto_model.layers.22.mlp.act_fn\n",
|
|
"0.auto_model.layers.22.input_layernorm\n",
|
|
"0.auto_model.layers.22.post_attention_layernorm\n",
|
|
"0.auto_model.layers.22.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.22.post_feedforward_layernorm\n",
|
|
"0.auto_model.layers.23\n",
|
|
"0.auto_model.layers.23.self_attn\n",
|
|
"0.auto_model.layers.23.self_attn.q_proj\n",
|
|
"0.auto_model.layers.23.self_attn.k_proj\n",
|
|
"0.auto_model.layers.23.self_attn.v_proj\n",
|
|
"0.auto_model.layers.23.self_attn.o_proj\n",
|
|
"0.auto_model.layers.23.self_attn.q_norm\n",
|
|
"0.auto_model.layers.23.self_attn.k_norm\n",
|
|
"0.auto_model.layers.23.mlp\n",
|
|
"0.auto_model.layers.23.mlp.gate_proj\n",
|
|
"0.auto_model.layers.23.mlp.up_proj\n",
|
|
"0.auto_model.layers.23.mlp.down_proj\n",
|
|
"0.auto_model.layers.23.mlp.act_fn\n",
|
|
"0.auto_model.layers.23.input_layernorm\n",
|
|
"0.auto_model.layers.23.post_attention_layernorm\n",
|
|
"0.auto_model.layers.23.pre_feedforward_layernorm\n",
|
|
"0.auto_model.layers.23.post_feedforward_layernorm\n",
|
|
"0.auto_model.norm\n",
|
|
"0.auto_model.rotary_emb\n",
|
|
"0.auto_model.rotary_emb_local\n",
|
|
"1\n",
|
|
"2\n",
|
|
"2.activation_function\n",
|
|
"2.linear\n",
|
|
"3\n",
|
|
"3.activation_function\n",
|
|
"3.linear\n",
|
|
"4\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"for name, module in model.named_modules():\n",
|
|
" print(name)\n",
|
|
"\n",
|
|
"\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"id": "de6a257d",
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": []
|
|
}
|
|
],
|
|
"metadata": {
|
|
"kernelspec": {
|
|
"display_name": ".venv",
|
|
"language": "python",
|
|
"name": "python3"
|
|
},
|
|
"language_info": {
|
|
"codemirror_mode": {
|
|
"name": "ipython",
|
|
"version": 3
|
|
},
|
|
"file_extension": ".py",
|
|
"mimetype": "text/x-python",
|
|
"name": "python",
|
|
"nbconvert_exporter": "python",
|
|
"pygments_lexer": "ipython3",
|
|
"version": "3.10.12"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 5
|
|
}
|