updated with new models OpenHermes and BAAI/bge-large embedding model

.env

@@ -4,8 +4,8 @@ ENVIRONMENT=dev
 DB_HOST=localhost
 DB_USER=postgres
 DB_PORT=5432
-DB_PASSWORD=quick
-DB_NAME=QuickGpt
+DB_PASSWORD=admin
+DB_NAME=openai
 
 SUPER_ADMIN_EMAIL=superadmin@email.com
 SUPER_ADMIN_PASSWORD=supersecretpassword

.gitignore

@@ -36,3 +36,4 @@ __pycache__/
 *.docx
 
 *.png
+

local_data/.gitignore

@@ -0,0 +1,2 @@
+*
+!.gitignore

local_data/private_gpt/graph_store.json

@@ -1 +0,0 @@
-{"graph_dict": {}}

local_data/private_gpt/image__vector_store.json

@@ -1 +0,0 @@
-{"embedding_dict": {}, "text_id_to_ref_doc_id": {}, "metadata_dict": {}}

local_data/private_gpt/qdrant/.lock

@@ -1 +0,0 @@
-tmp lock file

local_data/private_gpt/qdrant/meta.json

@@ -1 +1 @@
-{"collections": {"make_this_parameterizable_per_api_call": {"vectors": {"size": 768, "distance": "Cosine", "hnsw_config": null, "quantization_config": null, "on_disk": null}, "shard_number": null, "sharding_method": null, "replication_factor": null, "write_consistency_factor": null, "on_disk_payload": null, "hnsw_config": null, "wal_config": null, "optimizers_config": null, "init_from": null, "quantization_config": null, "sparse_vectors": null}}, "aliases": {}}
+{"collections": {"make_this_parameterizable_per_api_call": {"vectors": {"size": 1024, "distance": "Cosine", "hnsw_config": null, "quantization_config": null, "on_disk": null}, "shard_number": null, "sharding_method": null, "replication_factor": null, "write_consistency_factor": null, "on_disk_payload": null, "hnsw_config": null, "wal_config": null, "optimizers_config": null, "init_from": null, "quantization_config": null, "sparse_vectors": null}}, "aliases": {}}

models/.gitignore

@@ -0,0 +1,2 @@
+*
+!.gitignore

private_gpt/server/ingest/ingest_router.py

@@ -211,6 +211,7 @@ async def create_documents(
     )
     print("DOCUMENT CREATE: ", docs_in)
     document = crud.documents.create(db=db, obj_in=docs_in)
+    department_ids = department_ids if department_ids else "1"
     department_ids = [int(number) for number in department_ids.split(",")]
     for department_id in department_ids:
         db.execute(models.document_department_association.insert().values(document_id=document.id, department_id=department_id))

private_gpt/users/api/v1/routers/users.py

@@ -161,7 +161,8 @@ def read_user_me(
         username=current_user.username,
         company_id = current_user.company_id,
         department_id=current_user.department_id,
-        role =role
+        role =role,
+        checker=current_user.checker
     )
     return JSONResponse(
         status_code=status.HTTP_200_OK,

settings.yaml

@@ -41,14 +41,16 @@ llm:
   max_new_tokens: 512
   context_window: 3900
   temperature: 0.1      # The temperature of the model. Increasing the temperature will make the model answer more creatively. A value of 0.1 would be more factual. (Default: 0.1)
+  tokenizer: mistralai/Mistral-7B-Instruct-v0.2
   
 llamacpp:
-  prompt_style: "mistral"
-  llm_hf_repo_id: TheBloke/Mistral-7B-Instruct-v0.2-GGUF
-  llm_hf_model_file: mistral-7b-instruct-v0.2.Q4_K_M.gguf
+  prompt_style: "chatml"
+  llm_hf_repo_id: TheBloke/OpenHermes-2.5-Mistral-7B-GGUF
+  llm_hf_model_file: openhermes-2.5-mistral-7b.Q5_K_M.gguf
   tfs_z: 1.0            # Tail free sampling is used to reduce the impact of less probable tokens from the output. A higher value (e.g., 2.0) will reduce the impact more, while a value of 1.0 disables this setting
   top_k: 40             # Reduces the probability of generating nonsense. A higher value (e.g. 100) will give more diverse answers, while a lower value (e.g. 10) will be more conservative. (Default: 40)
-  top_p: 1.0            # Works together with top-k. A higher value (e.g., 0.95) will lead to more diverse text, while a lower value (e.g., 0.5) will generate more focused and conservative text. (Default: 0.9)
+  top_p: 0.9            # Works together with top-k. A higher value (e.g., 0.95) will lead to more diverse text, while a lower value (e.g., 0.5) will generate more focused and conservative text. (Default: 0.9)
+  repeat_last_n: 64     # Sets how far back for the model to look back to prevent repetition. (Default: 64, 0 = disabled, -1 = num_ctx)
   repeat_penalty: 1.1   # Sets how strongly to penalize repetitions. A higher value (e.g., 1.5) will penalize repetitions more strongly, while a lower value (e.g., 0.9) will be more lenient. (Default: 1.1)
 
 embedding:
@@ -57,7 +59,7 @@ embedding:
   ingest_mode: simple
 
 huggingface:
-  embedding_hf_model_name: BAAI/bge-small-en-v1.5
+  embedding_hf_model_name: BAAI/bge-large-en-v1.5
 
 vectorstore:
   database: qdrant
@@ -71,7 +73,7 @@ pgvector:
   database: postgres
   user: postgres
   password: postgres
-  embed_dim: 384 # 384 is for BAAI/bge-small-en-v1.5
+  embed_dim: 768 # 384 is for BAAI/bge-small-en-v1.5
   schema_name: private_gpt
   table_name: embeddings