rag/rag_fr.ipynb
2024-01-03 08:46:11 +01:00

1121 lines
68 KiB
Plaintext
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"id": "6503d5b0-6cf3-42d8-982c-353eb42d9d26",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/Users/peportier/miniforge3/envs/RAG_ENV/lib/python3.9/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
" from .autonotebook import tqdm as notebook_tqdm\n",
"/Users/peportier/miniforge3/envs/RAG_ENV/lib/python3.9/site-packages/transformers/utils/generic.py:441: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.\n",
" _torch_pytree._register_pytree_node(\n"
]
}
],
"source": [
"from llama_index import (\n",
" SimpleDirectoryReader,\n",
" VectorStoreIndex,\n",
" ServiceContext,\n",
" set_global_tokenizer,\n",
" load_index_from_storage,\n",
")\n",
"from llama_index.llms import LlamaCPP\n",
"from llama_index.llms.llama_utils import (\n",
" messages_to_prompt,\n",
" completion_to_prompt,\n",
")\n",
"from llama_index.vector_stores import ChromaVectorStore\n",
"from llama_index.storage.storage_context import StorageContext\n",
"from llama_index.embeddings import HuggingFaceEmbedding\n",
"from llama_index.query_engine import CitationQueryEngine\n",
"from llama_index.prompts import PromptTemplate\n",
"\n",
"from IPython.display import Markdown, display\n",
"\n",
"from transformers import AutoTokenizer\n",
"\n",
"import os\n",
"\n",
"import chromadb"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "15e2462f-5d5b-4f4b-8119-de782c400d78",
"metadata": {},
"outputs": [],
"source": [
"def messages_to_prompt(messages):\n",
" prompt = \"\"\n",
" for message in messages:\n",
" if message.role == 'system':\n",
" prompt += f\"<|system|>\\n{message.content}</s>\\n\"\n",
" elif message.role == 'user':\n",
" prompt += f\"<|user|>\\n{message.content}</s>\\n\"\n",
" elif message.role == 'assistant':\n",
" prompt += f\"<|assistant|>\\n{message.content}</s>\\n\"\n",
"\n",
" # ensure we start with a system prompt, insert blank if needed\n",
" if not prompt.startswith(\"<|system|>\\n\"):\n",
" prompt = \"<|system|>\\n</s>\\n\" + prompt\n",
"\n",
" # add final assistant prompt\n",
" prompt = prompt + \"<|assistant|>\\n\"\n",
"\n",
" return prompt"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "373e13fb-9233-4b0b-8dc3-a1e24117bd76",
"metadata": {},
"outputs": [],
"source": [
"def completion_to_prompt(completion, system_prompt=None):\n",
" prompt = \"\"\n",
" system_prompt_str = system_prompt or \"\"\"\\\n",
"Vous fournissez avec soin des réponses précises, factuelles, réfléchies et nuancées, et vous êtes doué pour le raisonnement. \\\n",
"Si vous pensez qu'il n'y a peut-être pas de bonne réponse, vous le dites. \\\n",
"Ne soyez pas verbeux dans vos réponses, mais donnez des détails et des exemples lorsque cela peut aider à l'explication. \\\n",
"\"\"\"\n",
" prompt = f\"<|system|>\\n {system_prompt_str.strip()} </s>\\n\"\n",
" prompt += f\"<|user|>\\n {completion} </s>\\n\"\n",
" prompt += f\"<|assistant|>\\n\"\n",
" return prompt"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "05f54aec-f155-4f4b-9b19-874c13db87cc",
"metadata": {
"collapsed": true,
"jupyter": {
"outputs_hidden": true
}
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"llama_model_loader: loaded meta data with 21 key-value pairs and 291 tensors from /Users/peportier/llm/a/a/zephyr-7b-beta.Q5_K_M.gguf (version GGUF V3 (latest))\n",
"llama_model_loader: - tensor 0: token_embd.weight q5_K [ 4096, 32000, 1, 1 ]\n",
"llama_model_loader: - tensor 1: blk.0.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 2: blk.0.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 3: blk.0.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 4: blk.0.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 5: blk.0.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 6: blk.0.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 7: blk.0.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 8: blk.0.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 9: blk.0.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 10: blk.1.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 11: blk.1.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 12: blk.1.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 13: blk.1.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 14: blk.1.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 15: blk.1.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 16: blk.1.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 17: blk.1.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 18: blk.1.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 19: blk.2.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 20: blk.2.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 21: blk.2.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 22: blk.2.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 23: blk.2.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 24: blk.2.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 25: blk.2.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 26: blk.2.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 27: blk.2.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 28: blk.3.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 29: blk.3.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 30: blk.3.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 31: blk.3.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 32: blk.3.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 33: blk.3.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 34: blk.3.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 35: blk.3.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 36: blk.3.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 37: blk.4.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 38: blk.4.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 39: blk.4.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 40: blk.4.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 41: blk.4.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 42: blk.4.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 43: blk.4.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 44: blk.4.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 45: blk.4.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 46: blk.5.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 47: blk.5.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 48: blk.5.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 49: blk.5.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 50: blk.5.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 51: blk.5.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 52: blk.5.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 53: blk.5.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 54: blk.5.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 55: blk.6.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 56: blk.6.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 57: blk.6.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 58: blk.6.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 59: blk.6.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 60: blk.6.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 61: blk.6.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 62: blk.6.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 63: blk.6.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 64: blk.7.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 65: blk.7.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 66: blk.7.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 67: blk.7.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 68: blk.7.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 69: blk.7.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 70: blk.7.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 71: blk.7.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 72: blk.7.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 73: blk.8.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 74: blk.8.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 75: blk.8.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 76: blk.8.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 77: blk.10.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 78: blk.10.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 79: blk.10.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 80: blk.10.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 81: blk.10.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 82: blk.10.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 83: blk.10.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 84: blk.10.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 85: blk.10.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 86: blk.11.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 87: blk.11.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 88: blk.11.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 89: blk.11.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 90: blk.11.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 91: blk.11.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 92: blk.11.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 93: blk.11.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 94: blk.11.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 95: blk.12.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 96: blk.12.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 97: blk.12.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 98: blk.12.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 99: blk.12.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 100: blk.12.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 101: blk.8.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 102: blk.8.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 103: blk.8.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 104: blk.8.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 105: blk.8.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 106: blk.9.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 107: blk.9.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 108: blk.9.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 109: blk.9.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 110: blk.9.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 111: blk.9.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 112: blk.9.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 113: blk.9.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 114: blk.9.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 115: blk.12.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 116: blk.12.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 117: blk.12.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 118: blk.13.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 119: blk.13.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 120: blk.13.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 121: blk.13.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 122: blk.13.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 123: blk.13.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 124: blk.13.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 125: blk.13.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 126: blk.13.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 127: blk.14.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 128: blk.14.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 129: blk.14.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 130: blk.14.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 131: blk.14.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 132: blk.14.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 133: blk.14.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 134: blk.14.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 135: blk.14.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 136: blk.15.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 137: blk.15.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 138: blk.15.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 139: blk.15.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 140: blk.15.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 141: blk.15.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 142: blk.15.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 143: blk.15.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 144: blk.15.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 145: blk.16.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 146: blk.16.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 147: blk.16.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 148: blk.16.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 149: blk.16.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 150: blk.16.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 151: blk.16.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 152: blk.16.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 153: blk.16.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 154: blk.17.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 155: blk.17.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 156: blk.17.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 157: blk.17.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 158: blk.17.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 159: blk.17.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 160: blk.17.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 161: blk.17.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 162: blk.17.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 163: blk.18.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 164: blk.18.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 165: blk.18.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 166: blk.18.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 167: blk.18.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 168: blk.18.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 169: blk.18.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 170: blk.18.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 171: blk.18.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 172: blk.19.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 173: blk.19.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 174: blk.19.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 175: blk.19.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 176: blk.19.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 177: blk.19.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 178: blk.19.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 179: blk.19.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 180: blk.19.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 181: blk.20.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 182: blk.20.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 183: blk.20.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 184: blk.20.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 185: blk.20.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 186: blk.20.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 187: blk.20.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 188: blk.20.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 189: blk.20.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 190: blk.21.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 191: blk.21.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 192: blk.21.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 193: blk.21.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 194: blk.21.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 195: blk.21.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 196: blk.21.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 197: blk.21.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 198: blk.21.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 199: blk.22.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 200: blk.22.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 201: blk.22.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 202: blk.22.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 203: blk.22.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 204: blk.22.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 205: blk.22.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 206: blk.22.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 207: blk.22.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 208: blk.23.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 209: blk.23.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 210: blk.23.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 211: blk.23.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 212: blk.23.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 213: blk.23.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 214: blk.23.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 215: blk.23.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 216: blk.23.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 217: blk.24.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 218: blk.24.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 219: blk.24.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 220: blk.24.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 221: blk.24.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 222: blk.24.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 223: blk.24.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 224: blk.24.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 225: blk.24.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 226: blk.25.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 227: blk.25.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 228: blk.25.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 229: blk.25.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 230: blk.25.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 231: blk.25.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 232: blk.25.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 233: blk.25.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 234: blk.25.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 235: blk.26.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 236: blk.26.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 237: blk.26.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 238: blk.26.attn_v.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 239: blk.26.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 240: blk.26.ffn_down.weight q5_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 241: blk.26.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 242: blk.26.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 243: blk.26.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 244: blk.27.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 245: blk.27.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 246: blk.27.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 247: blk.27.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 248: blk.27.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 249: blk.27.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 250: blk.27.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 251: blk.27.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 252: blk.27.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 253: blk.28.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 254: blk.28.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 255: blk.28.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 256: blk.28.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 257: blk.28.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 258: blk.28.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 259: blk.28.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 260: blk.28.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 261: blk.28.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 262: blk.29.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 263: blk.29.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 264: blk.29.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 265: blk.29.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 266: blk.29.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 267: blk.29.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 268: blk.29.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 269: blk.29.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 270: blk.29.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 271: blk.30.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 272: blk.30.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 273: blk.30.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 274: blk.30.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 275: blk.30.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 276: blk.30.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 277: output.weight q6_K [ 4096, 32000, 1, 1 ]\n",
"llama_model_loader: - tensor 278: blk.30.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 279: blk.30.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 280: blk.30.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 281: blk.31.attn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 282: blk.31.ffn_down.weight q6_K [ 14336, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 283: blk.31.ffn_gate.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 284: blk.31.ffn_up.weight q5_K [ 4096, 14336, 1, 1 ]\n",
"llama_model_loader: - tensor 285: blk.31.ffn_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - tensor 286: blk.31.attn_k.weight q5_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 287: blk.31.attn_output.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 288: blk.31.attn_q.weight q5_K [ 4096, 4096, 1, 1 ]\n",
"llama_model_loader: - tensor 289: blk.31.attn_v.weight q6_K [ 4096, 1024, 1, 1 ]\n",
"llama_model_loader: - tensor 290: output_norm.weight f32 [ 4096, 1, 1, 1 ]\n",
"llama_model_loader: - kv 0: general.architecture str = llama\n",
"llama_model_loader: - kv 1: general.name str = huggingfaceh4_zephyr-7b-beta\n",
"llama_model_loader: - kv 2: llama.context_length u32 = 32768\n",
"llama_model_loader: - kv 3: llama.embedding_length u32 = 4096\n",
"llama_model_loader: - kv 4: llama.block_count u32 = 32\n",
"llama_model_loader: - kv 5: llama.feed_forward_length u32 = 14336\n",
"llama_model_loader: - kv 6: llama.rope.dimension_count u32 = 128\n",
"llama_model_loader: - kv 7: llama.attention.head_count u32 = 32\n",
"llama_model_loader: - kv 8: llama.attention.head_count_kv u32 = 8\n",
"llama_model_loader: - kv 9: llama.attention.layer_norm_rms_epsilon f32 = 0.000010\n",
"llama_model_loader: - kv 10: llama.rope.freq_base f32 = 10000.000000\n",
"llama_model_loader: - kv 11: general.file_type u32 = 17\n",
"llama_model_loader: - kv 12: tokenizer.ggml.model str = llama\n",
"llama_model_loader: - kv 13: tokenizer.ggml.tokens arr[str,32000] = [\"<unk>\", \"<s>\", \"</s>\", \"<0x00>\", \"<...\n",
"llama_model_loader: - kv 14: tokenizer.ggml.scores arr[f32,32000] = [0.000000, 0.000000, 0.000000, 0.0000...\n",
"llama_model_loader: - kv 15: tokenizer.ggml.token_type arr[i32,32000] = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...\n",
"llama_model_loader: - kv 16: tokenizer.ggml.bos_token_id u32 = 1\n",
"llama_model_loader: - kv 17: tokenizer.ggml.eos_token_id u32 = 2\n",
"llama_model_loader: - kv 18: tokenizer.ggml.unknown_token_id u32 = 0\n",
"llama_model_loader: - kv 19: tokenizer.ggml.padding_token_id u32 = 2\n",
"llama_model_loader: - kv 20: general.quantization_version u32 = 2\n",
"llama_model_loader: - type f32: 65 tensors\n",
"llama_model_loader: - type q5_K: 193 tensors\n",
"llama_model_loader: - type q6_K: 33 tensors\n",
"llm_load_vocab: special tokens definition check successful ( 259/32000 ).\n",
"llm_load_print_meta: format = GGUF V3 (latest)\n",
"llm_load_print_meta: arch = llama\n",
"llm_load_print_meta: vocab type = SPM\n",
"llm_load_print_meta: n_vocab = 32000\n",
"llm_load_print_meta: n_merges = 0\n",
"llm_load_print_meta: n_ctx_train = 32768\n",
"llm_load_print_meta: n_embd = 4096\n",
"llm_load_print_meta: n_head = 32\n",
"llm_load_print_meta: n_head_kv = 8\n",
"llm_load_print_meta: n_layer = 32\n",
"llm_load_print_meta: n_rot = 128\n",
"llm_load_print_meta: n_gqa = 4\n",
"llm_load_print_meta: f_norm_eps = 0.0e+00\n",
"llm_load_print_meta: f_norm_rms_eps = 1.0e-05\n",
"llm_load_print_meta: f_clamp_kqv = 0.0e+00\n",
"llm_load_print_meta: f_max_alibi_bias = 0.0e+00\n",
"llm_load_print_meta: n_ff = 14336\n",
"llm_load_print_meta: rope scaling = linear\n",
"llm_load_print_meta: freq_base_train = 10000.0\n",
"llm_load_print_meta: freq_scale_train = 1\n",
"llm_load_print_meta: n_yarn_orig_ctx = 32768\n",
"llm_load_print_meta: rope_finetuned = unknown\n",
"llm_load_print_meta: model type = 7B\n",
"llm_load_print_meta: model ftype = mostly Q5_K - Medium\n",
"llm_load_print_meta: model params = 7.24 B\n",
"llm_load_print_meta: model size = 4.78 GiB (5.67 BPW) \n",
"llm_load_print_meta: general.name = huggingfaceh4_zephyr-7b-beta\n",
"llm_load_print_meta: BOS token = 1 '<s>'\n",
"llm_load_print_meta: EOS token = 2 '</s>'\n",
"llm_load_print_meta: UNK token = 0 '<unk>'\n",
"llm_load_print_meta: PAD token = 2 '</s>'\n",
"llm_load_print_meta: LF token = 13 '<0x0A>'\n",
"llm_load_tensors: ggml ctx size = 0.11 MiB\n",
"llm_load_tensors: mem required = 4893.10 MiB\n",
"...................................................................................................\n",
"llama_new_context_with_model: n_ctx = 3900\n",
"llama_new_context_with_model: freq_base = 10000.0\n",
"llama_new_context_with_model: freq_scale = 1\n",
"llama_new_context_with_model: kv self size = 487.50 MiB\n",
"llama_build_graph: non-view tensors processed: 740/740\n",
"ggml_metal_init: allocating\n",
"ggml_metal_init: found device: Apple M2 Max\n",
"ggml_metal_init: picking default device: Apple M2 Max\n",
"ggml_metal_init: default.metallib not found, loading from source\n",
"ggml_metal_init: loading '/Users/peportier/miniforge3/envs/RAG_ENV/lib/python3.9/site-packages/llama_cpp/ggml-metal.metal'\n",
"ggml_metal_init: GPU name: Apple M2 Max\n",
"ggml_metal_init: GPU family: MTLGPUFamilyApple8 (1008)\n",
"ggml_metal_init: hasUnifiedMemory = true\n",
"ggml_metal_init: recommendedMaxWorkingSetSize = 49152.00 MiB\n",
"ggml_metal_init: maxTransferRate = built-in GPU\n",
"llama_new_context_with_model: compute buffer total size = 278.43 MiB\n",
"llama_new_context_with_model: max tensor size = 102.54 MiB\n",
"ggml_metal_add_buffer: allocated 'data ' buffer, size = 4893.70 MiB, ( 4894.33 / 49152.00)\n",
"ggml_metal_add_buffer: allocated 'kv ' buffer, size = 487.52 MiB, ( 5381.84 / 49152.00)\n",
"ggml_metal_add_buffer: allocated 'alloc ' buffer, size = 275.38 MiB, ( 5657.22 / 49152.00)\n"
]
}
],
"source": [
"llm = LlamaCPP(\n",
" # You can pass in the URL to a GGML model to download it automatically\n",
" model_url=None,\n",
" # optionally, you can set the path to a pre-downloaded model instead of model_url\n",
" # model_path='/Users/peportier/llm/a/a/mistral-7b-openorca.Q4_K_M.gguf',\n",
" model_path='/Users/peportier/llm/a/a/zephyr-7b-beta.Q5_K_M.gguf',\n",
" temperature=0.1,\n",
" max_new_tokens=1024,\n",
" # llama2 has a context window of 4096 tokens, but we set it lower to allow for some wiggle room\n",
" context_window=3900,\n",
" # kwargs to pass to __call__()\n",
" # https://llama-cpp-python.readthedocs.io/en/latest/api-reference/#llama_cpp.Llama.__call__\n",
" generate_kwargs={\n",
" \"temperature\": 0.1,\n",
" \"mirostat_mode\": 2,\n",
" },\n",
" # kwargs to pass to __init__()\n",
" # https://llama-cpp-python.readthedocs.io/en/latest/api-reference/#llama_cpp.Llama.__init__\n",
" # set to at least 1 to use GPU\n",
" model_kwargs={\n",
" \"n_gpu_layers\": 1,\n",
" },\n",
" # transform inputs into Llama2 format\n",
" messages_to_prompt=messages_to_prompt,\n",
" completion_to_prompt=completion_to_prompt,\n",
" verbose=False,\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "42734468-f0c4-46ff-8676-98d4bde99a86",
"metadata": {},
"outputs": [],
"source": [
"model_name = \"HuggingFaceH4/zephyr-7b-beta\"\n",
"set_global_tokenizer(\n",
" AutoTokenizer.from_pretrained(model_name).encode\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "39fadac2-a30e-41ca-85b5-35fc49f9842c",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/Users/peportier/miniforge3/envs/RAG_ENV/lib/python3.9/site-packages/transformers/utils/generic.py:309: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.\n",
" _torch_pytree._register_pytree_node(\n"
]
}
],
"source": [
"# embed_model = HuggingFaceEmbedding(model_name=\"BAAI/bge-small-en-v1.5\")\n",
"# embed_model = HuggingFaceEmbedding(model_name=\"sentence-transformers/all-MiniLM-L6-v2\")\n",
"embed_model = HuggingFaceEmbedding(model_name=\"sentence-transformers/distiluse-base-multilingual-cased-v1\")"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "fbb8f6f3-9175-4f90-bd67-be8efb37ae17",
"metadata": {},
"outputs": [],
"source": [
"service_context = ServiceContext.from_defaults(\n",
" llm=llm,\n",
" embed_model=embed_model,\n",
" chunk_size=512,\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "c06f2c57-8f7b-41ad-9053-bdfefc80fd98",
"metadata": {},
"outputs": [],
"source": [
"if not os.path.exists(\"./index_cera1\"):\n",
" documents = SimpleDirectoryReader(\"./docs/cera1\").load_data()\n",
" index = VectorStoreIndex.from_documents(\n",
" documents, service_context=service_context\n",
" )\n",
" index.storage_context.persist(persist_dir=\"./index_cera1\")\n",
"else:\n",
" index = load_index_from_storage(\n",
" StorageContext.from_defaults(persist_dir=\"./index_cera1\"),\n",
" service_context=service_context,\n",
" )"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "b583c0c0-85a1-4384-83de-0001c9452a3f",
"metadata": {},
"outputs": [],
"source": [
"# db = chromadb.PersistentClient(path=\"./chroma_db\")\n",
"# chroma_collection = db.get_or_create_collection(\"env\")\n",
"# vector_store = ChromaVectorStore(chroma_collection=chroma_collection)\n",
"# storage_context = StorageContext.from_defaults(vector_store=vector_store)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "a7a6a9f9-a3d6-47fe-8cd8-13384e07f16a",
"metadata": {},
"outputs": [],
"source": [
"# # Create chroma index\n",
"# index = VectorStoreIndex.from_documents(\n",
"# documents, storage_context=storage_context,\n",
"# service_context=service_context\n",
"# )"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "176a1285-bb3c-46a4-892a-1957ccb74c99",
"metadata": {},
"outputs": [],
"source": [
"text_qa_template_str_fr = (\n",
" \"<|system|>: Vous êtes un assistant IA qui répond à la question posée à la fin en utilisant le contexte suivant. Toutes les informations factuelles que vous utilisez pour répondre proviennent exclusivement du contexte. Si vous ne connaissez pas la réponse, dites simplement que vous ne savez pas, n'essayez pas d'inventer une réponse. Veuillez répondre exclusivement en français. </s>\\n\"\n",
" \"<|user|>: {context_str}\\n\"\n",
" \"Question: {query_str} </s>\\n\"\n",
" \"<|assistant|>:\"\n",
")\n",
"\n",
"text_qa_template = PromptTemplate(text_qa_template_str_fr)\n",
"\n",
"refine_template_str_fr = (\n",
" \"<|user|>: La requête originale est la suivante : {query_str}\\n\"\n",
" \"Nous avons fourni une première réponse : {existing_answer}\\n\"\n",
" \"Nous avons la possibilité d'affiner la réponse existante (seulement si nécessaire) avec un peu plus de contexte ci-dessous.\\n\"\n",
" \"------------\\n\"\n",
" \"{context_msg}\\n\"\n",
" \"------------\\n\"\n",
" \"Compte tenu du nouveau contexte, la réponse initiale est affinée afin de mieux répondre à la requête. Si le contexte n'est pas utile, renvoyer la réponse originale. </s>\\n\"\n",
" \"<|assistant|>:\"\n",
")\n",
"refine_template = PromptTemplate(refine_template_str_fr)"
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "02350391-a152-4b70-8ed4-a8a3dde96728",
"metadata": {},
"outputs": [],
"source": [
"query_engine = index.as_query_engine(\n",
" text_qa_template=text_qa_template,\n",
" refine_template=refine_template,\n",
" #response_mode=\"compact\",\n",
" response_mode=\"refine\",\n",
" similarity_top_k=5,\n",
" streaming=True)"
]
},
{
"cell_type": "code",
"execution_count": 12,
"id": "f979bd65-217d-480b-a377-5b3a58593156",
"metadata": {},
"outputs": [],
"source": [
"def display_prompt_dict(prompts_dict):\n",
" for k, p in prompts_dict.items():\n",
" text_md = f\"**Prompt Key**: {k}<br>\" f\"**Text:** <br>\"\n",
" display(Markdown(text_md))\n",
" print(p.get_template())\n",
" display(Markdown(\"<br><br>\"))"
]
},
{
"cell_type": "code",
"execution_count": 13,
"id": "37ba2a5d-5eca-43aa-ac28-b071e1cbfba4",
"metadata": {},
"outputs": [
{
"data": {
"text/markdown": [
"**Prompt Key**: response_synthesizer:text_qa_template<br>**Text:** <br>"
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"<|system|>: Vous êtes un assistant IA qui répond à la question posée à la fin en utilisant le contexte suivant. Toutes les informations factuelles que vous utilisez pour répondre proviennent exclusivement du contexte. Si vous ne connaissez pas la réponse, dites simplement que vous ne savez pas, n'essayez pas d'inventer une réponse. Veuillez répondre exclusivement en français. </s>\n",
"<|user|>: {context_str}\n",
"Question: {query_str} </s>\n",
"<|assistant|>:\n"
]
},
{
"data": {
"text/markdown": [
"<br><br>"
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/markdown": [
"**Prompt Key**: response_synthesizer:refine_template<br>**Text:** <br>"
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"<|user|>: La requête originale est la suivante : {query_str}\n",
"Nous avons fourni une première réponse : {existing_answer}\n",
"Nous avons la possibilité d'affiner la réponse existante (seulement si nécessaire) avec un peu plus de contexte ci-dessous.\n",
"------------\n",
"{context_msg}\n",
"------------\n",
"Compte tenu du nouveau contexte, la réponse initiale est affinée afin de mieux répondre à la requête. Si le contexte n'est pas utile, renvoyer la réponse originale. </s>\n",
"<|assistant|>:\n"
]
},
{
"data": {
"text/markdown": [
"<br><br>"
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"display_prompt_dict(query_engine.get_prompts())"
]
},
{
"cell_type": "code",
"execution_count": 27,
"id": "8bcff0b7-300b-4698-86df-f36171b1ee02",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"En complément de notre réponse initiale, la Caisse d'Epargne Rhône Alpes a récemment financé un projet ambitieux de construction de 44 centrales solaires pour le compte de Terr e et Lac, une société spécialisée dans la production d'énergies renouvelables. Le montant total de ce financement bancaire s'élève à 14,5 millions d'euros et permettra la construction de centrales photovoltaïques sur les toitures, ombrières de parking et sol principalement situées en Auvergne-Rhône-Alpes. Ces installations représentent une capacité totale de 12,5 MWc. La Caisse d'Epargne Rhône Alpes est une banque régionale coopérative qui joue un rôle prépondérant dans le financement de la transition environnementale sur son territoire. Elle accompagne les entreprises de toutes tailles, les projets structurants des collectivités territoriales et le développement des producteurs d'énergies renouvelables en vertu de son engagement au Contrat d'Utilité. Les clients particuliers avec des projets de rénovation bénéficient des conseils de France Rénov, partenaire de la Caisse dEpargne Rhône Alpes. Pour les entreprises et collectivités, la transition énergétique est un enjeu majeur. La Caisse d'Epargne Rhône Alpes accompagne leur démarche de transformation via des financements adaptés, comme le Prêt à Impact - une offre innovante de financement dont le taux dintérêt est indexé sur des objectifs environnementaux et/ou sociétaux - ou encore une gamme de prêts « green » à disposition des acteurs économiques pour leur transition écologique. De plus, elle propose des conseils et expertises sur-mesure selon les besoins et secteurs dactivité.\n",
"\n",
"Ce financement s'inscrit dans la feuille de route de la Caisse d'Epargne Rhône Alpes qui a mis au cœur de ses priorités la transition énergétique et le financement des ENR à l'échelle de ses territoires. Selon Sébastien Fenet, Directeur Général de Terre et Lac : « Depuis sa création, Terre et Lac accompagne les acteurs de notre région, entreprises, agriculteurs et collectivités locales dans la transition énergétique. Forts de cet ancrage territorial que n'ous sommes fiers de partager avec notre partenaire bancaire la Caisse dEpargne Rhône Alpes , nous contribuons ensemble au développement des énergies renouvelables avec la même préoccupation : développer des projets respectueux de leur environnement en étroite collaboration avec le sénat de notre territoire ».\n",
"\n",
"La Caisse d'Epargne Rhône Alpes est une banque régionale, commerciale et coopérative de plein exercice présente sur cinq départements (Ain, Isère, Rhône, Savoie et Haute-Savoie) et sur tous les métiers de la banque. Elle compte 1,4 million de clients, 458 000 sociétaires, 3 000 collaborateurs et 280 agences.CPU times: user 8.12 s, sys: 932 ms, total: 9.05 s\n",
"Wall time: 1min 26s\n"
]
}
],
"source": [
"%%time\n",
"response = query_engine.query('''Quels sont les offres commerciales de la Caisse d'Epargne Rhône Alpes en faveur de la transition environnementale ?''')\n",
"response.print_response_stream()"
]
},
{
"cell_type": "code",
"execution_count": 17,
"id": "7c7f1e8b-cb21-4cc6-8d51-8e29ced1ba6a",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"88c623578b.txt\n",
"bdc7145c09.pdf\n",
"7a4caa88da.pdf\n",
"0a235f3915.txt\n",
"bdc7145c09.pdf\n"
]
}
],
"source": [
"for node in response.source_nodes:\n",
" print(node.metadata[\"file_name\"])"
]
},
{
"cell_type": "code",
"execution_count": 28,
"id": "876154c4-2dbb-4170-a244-98f17e763955",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"88c623578b.txt\n",
"\n",
"---------------------------\n",
"\n"
]
},
{
"data": {
"text/markdown": [
"caissedepargnerhonealpes.fr/responsable-solidaire/)\n",
"Développement durable\n",
"\n",
"Partager\n",
"\n",
"# Engagée dans le Développement Durable\n",
"\n",
"Agir pour la transition écologique\n",
"\n",
"Réduire notre empreinte environnementale\n",
"\n",
"Dans le cadre de sa démarche RSE, la Caisse dEpargne Rhône Alpes poursuit une\n",
"politique volontariste pour répondre aux défis de la transition énergétique.\n",
"Elle accompagne lensemble de ses clients dans la transition vers une économie\n",
"plus durable, et agit à son niveau pour limiter limpact de son activité sur\n",
"lenvironnement.\n",
"\n",
"## **Agir pour la transition écologique**\n",
"\n",
"![](https://www.caissedepargnerhonealpes.fr/wp-\n",
"content/uploads/2021/11/491843054-2-806x1024.jpg)\n",
"\n",
"**Accompagner la transformation des territoires**\n",
"\n",
"Acteur majeur de lEconomie Locale, la Caisse dEpargne Rhône Alpes joue\n",
"pleinement son rôle de grande banque coopérative régionale en favorisant le\n",
"financement dinfrastructures à impact positif sur notre territoire. Energies\n",
"renouvelables, ascenseurs valléens, rénovation énergétique de logements\n",
"sociaux, valorisation des déchets… elle accompagne de grands projets à haute\n",
"valeur environnementale."
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/markdown": [
"---"
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"efcbb5b260.txt\n",
"\n",
"---------------------------\n",
"\n"
]
},
{
"data": {
"text/markdown": [
"plus utile\n",
"\n",
"Contribuer à la transition énergétique\n",
"\n",
"Accompagner l'activité économique\n",
"\n",
"Soutenir les projets solidaires\n",
"\n",
"Favoriser l'emploi des jeunes\n",
"\n",
"Régionale et coopérative, la Caisse dEpargne Rhône Alpes sengage au plus\n",
"près des besoins du territoire pour accompagner les transitions économiques,\n",
"environnementales et sociales. Cest le Contrat dUtilité. Que représente-t-\n",
"il concrètement ? Voici quelques exemples de réalisation.\n",
"\n",
"## Être utile, c'est aussi…\n",
"\n",
"### **Cont** ribuer à la transition énergétique\n",
"\n",
"![](https://www.caissedepargnerhonealpes.fr/wp-\n",
"content/uploads/2023/09/CERA_AFF-600x800mm_4visuels_v2_00004-768x1024.png)\n",
"\n",
"La Caisse d'Epargne Rhône Alpes accompagne la **rénovation énergétique de plus\n",
"de 1 000 logements par an**. Les clients particuliers avec des projets de\n",
"rénovation bénéficient des [conseils](https://www.caisse-epargne.fr/rhone-\n",
"alpes/emprunter/maprimerenov/) de France Rénov, partenaire de la Caisse\n",
"dEpargne Rhône Alpes.\n",
"\n",
"**Pour les entreprises et collectivités, la transition énergétique est un\n",
"enjeu majeur**."
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/markdown": [
"---"
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"bdc7145c09.pdf\n",
"\n",
"---------------------------\n",
"\n"
]
},
{
"data": {
"text/markdown": [
"Communiqué de presse \n",
"Lyon, le 31 août 2023 \n",
" \n",
"La Caisse dEpargne Rhône Alpes finance un ambitieu x projet \n",
"dinstallations photovoltaïques de Terre et Lac \n",
" \n",
"La Caisse dEpargne Rhône Alpes arrange la mise en place dun financement bancaire de \n",
"14,5 millions deuros pour la construction par Terr e et Lac de 44 centrales solaires. \n",
" \n",
"Terre et Lac, producteur indépendant dénergies, es t un expert photovoltaïque des grandes \n",
"toitures, ombrières de parking et sol. La société a finalisé une opération de financement pour un \n",
"montant total de 14,5 M€ avec la Caisse dEpargne R hône Alpes. Cette opération permet la \n",
"constitution dun portefeuille de projets greenfield regroupant 44 centrales photovoltaïques en \n",
"toiture essentiellement situées en Auvergne-Rhône-A lpes. Ces installations représentent une \n",
"capacité totale de 12,5 MWc. \n",
" \n",
"La Caisse dEpargne Rhône Alpes, banque régionale c oopérative, est un acteur de premier plan \n",
"du financement de la transition environnementale su r son territoire. Forte de son ancrage \n",
"régional, elle accompagne la transition des entrepr ises de toutes tailles, les projets structurants \n",
"des collectivités territoriales et le développement des producteurs dénergies renouvelables."
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/markdown": [
"---"
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"efcbb5b260.txt\n",
"\n",
"---------------------------\n",
"\n"
]
},
{
"data": {
"text/markdown": [
"Les clients particuliers avec des projets de\n",
"rénovation bénéficient des [conseils](https://www.caisse-epargne.fr/rhone-\n",
"alpes/emprunter/maprimerenov/) de France Rénov, partenaire de la Caisse\n",
"dEpargne Rhône Alpes.\n",
"\n",
"**Pour les entreprises et collectivités, la transition énergétique est un\n",
"enjeu majeur**. La Caisse d'Epargne Rhône Alpes [accompagne leur démarche de\n",
"transformation](https://www.caisse-epargne.fr/rhone-\n",
"alpes/professionnels/conseils/choisir-solutions-durables-responsables/) via :\n",
"\n",
" * Des financements adaptés, comme le Prêt à Impact - une offre innovante de financement dont le taux dintérêt est indexé sur des objectifs environnementaux et/ou sociétaux - ou encore une gamme de prêts « green » à disposition des acteurs économiques pour leur transition écologique, \n",
" * Des conseils et expertises sur-mesure selon les besoins et secteurs dactivité."
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/markdown": [
"---"
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"bdc7145c09.pdf\n",
"\n",
"---------------------------\n",
"\n"
]
},
{
"data": {
"text/markdown": [
"Ce financement s inscrit dans la feuille de route de la Caisse \n",
"dEpargne Rhône Alpes qui a mis au cœur de ses prio rités la transition énergétique et le \n",
"financement des ENR à léchelle de ses territoires. » \n",
" \n",
"Sébastien Fenet , Directeur Général de Terre et Lac : « Depuis sa création, Terre et Lac accompagne \n",
"les acteurs de notre région, entreprises, agriculte urs et collectivité locales dans la transition \n",
"énergétique. Forts de cet ancrage territorial que n ous sommes fiers de partager avec notre \n",
"partenaire bancaire la Caisse dEpargne Rhône Alpes , nous contribuons ensemble au développement \n",
"des énergies renouvelables avec la même préoccupati on : développer des projets respectueux de \n",
"leur environnement en étroite collaboration avec le s habitants de notre territoire » \n",
" \n",
" \n",
"À propos de la Caisse dEpargne Rhône Alpes : \n",
"La Caisse dEpargne Rhône Alpes est une banque comm erciale, régionale et coopérative de plein exercice présente sur \n",
"cinq départements (Ain, Isère, Rhône, Savoie et Hau te-Savoie) et sur tous les métiers de la banque. El le agit au \n",
"quotidien pour le développement de son territoire. \n",
"Elle compte 1,4 million de clients, 458 000 sociéta ires, 3 000 collaborateurs, 280 agences."
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/markdown": [
"---"
],
"text/plain": [
"<IPython.core.display.Markdown object>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"for node in response.source_nodes:\n",
" print(node.metadata[\"file_name\"])\n",
" print(\"\\n---------------------------\\n\")\n",
" display(Markdown(node.text))\n",
" display(Markdown(\"---\"))"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "d42ad6a8-caae-46e9-96b6-bc8dba6397a7",
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "RAG_ENV",
"language": "python",
"name": "rag_env"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.9.18"
}
},
"nbformat": 4,
"nbformat_minor": 5
}