change clients

app.py

@@ -14,6 +14,13 @@ from backend.semantic_search import retrieve
 import itertools
 from gradio_client import Client
 
+
+clients_eps = {
+    "wizardlm 13B": "https://itsmynti-ehartford-wizardlm-13b-uncensored.hf.space/",
+    "mistral 7b v0.1": "https://zhuraavl-mistralai-mistral-7b-v0-1.hf.space/",
+}
+clients = {k: Client(u) for k,u in clients_eps.items()}
+
 client = Client("Be-Bo/llama-3-chatbot_70b")
 
 def run_llama(_, msg, *__):
@@ -22,13 +29,12 @@ def run_llama(_, msg, *__):
         api_name="/chat"
     )
 
-m7b01 = Client("https://zhuraavl-mistralai-mistral-7b-v0-1.hf.space/")
-def run_m7b01(_, msg, *__):
-    yield m7b01.predict(
+def run_client(cli_name, msg, *__):
+    yield clients[cli_name].predict(
         msg,
         api_name="/predict"
     )
-inf_models = list(hf_models.keys()) + list(openai_models)
+inf_models = list(hf_models.keys()) + list(openai_models) + list(clients_eps.keys())
 
 emb_models = ["bge", "minilm"]
 splitters = ['ct', 'rct', 'nltk']
@@ -99,8 +105,8 @@ def bot(history, model_name, oepnai_api_key,
 
     if model_name == "llama 3":
         generate_fn = run_llama
-    if model_name == "mistral-7B 0.1":
-        generate_fn = run_m7b01
+    if model_name in clients_eps:
+        generate_fn = run_client
     elif model_name in hf_models:
          generate_fn = generate_hf
     elif model_name in openai_models:

backend/query_llm.py

@@ -12,9 +12,6 @@ HF_TOKEN = os.getenv("HF_TOKEN")
 
 hf_models = {
     "mistral-7B": "mistralai/Mistral-7B-Instruct-v0.2",
-    "mistral-7B 0.1": "mistralai/Mistral-7B-v0.1",
-    # "vicuna-13b":"lmsys/vicuna-13b-v1.5",
-    # "WizardLM-30B": "cognitivecomputations/WizardLM-30B-Uncensored",
     "llama 3": "meta-llama/Meta-Llama-3-70B-Instruct",
 }
 openai_models = {"gpt-4o","gpt-3.5-turbo-0125"}