Use environment variables for selecting the models.

examples/Graph RAG

@@ -510,8 +510,7 @@
         "title": "Ask LLM",
         "params": {
           "max_tokens": 100.0,
-          "accepted_regex": "",
-          "model": "SultanR/SmolTulu-1.7b-Instruct"
+          "accepted_regex": ""
         },
         "display": null,
         "error": null,
@@ -541,13 +540,6 @@
               "type": {
                 "type": "<class 'int'>"
               }
-            },
-            "model": {
-              "type": {
-                "type": "<class 'str'>"
-              },
-              "default": null,
-              "name": "model"
             }
           },
           "outputs": {

lynxkite-lynxscribe/README.md

@@ -23,5 +23,9 @@ uv pip install infinity-emb[all]
 infinity_emb v2 --model-id michaelfeil/bge-small-en-v1.5
 uv pip install "sglang[all]>=0.4.2.post2" --find-links https://flashinfer.ai/whl/cu124/torch2.5/flashinfer/
 python -m sglang.launch_server --model-path SultanR/SmolTulu-1.7b-Instruct --port 8080
-LLM_BASE_URL='http://localhost:8080/v1' EMBEDDING_BASE_URL='http://localhost:7997/' lynxkite
+export LLM_BASE_URL='http://localhost:8080/v1'
+export LLM_MODEL='SultanR/SmolTulu-1.7b-Instruct'
+export EMBEDDING_BASE_URL='http://localhost:7997/'
+export EMBEDDING_MODEL='michaelfeil/bge-small-en-v1.5'
+lynxkite
 ```

lynxkite-lynxscribe/src/lynxkite_lynxscribe/llm_ops.py

@@ -130,8 +130,7 @@ def create_prompt(input, *, save_as="prompt", template: ops.LongStr):
 
 
 @op("Ask LLM")
-def ask_llm(input, *, model: str, accepted_regex: str = None, max_tokens: int = 100):
-    assert model, "Please specify the model."
+def ask_llm(input, *, accepted_regex: str = None, max_tokens: int = 100):
     assert "prompt" in input, "Please create the prompt first."
     options = {}
     if accepted_regex:
@@ -139,7 +138,6 @@ def ask_llm(input, *, model: str, accepted_regex: str = None, max_tokens: int =
             "regex": accepted_regex,
         }
     results = chat(
-        model=model,
         max_tokens=max_tokens,
         messages=[
             {"role": "user", "content": input["prompt"]},
@@ -228,10 +226,9 @@ def rag(
         results = [db[int(r)] for r in results["ids"][0]]
         return {**input, "rag": results, "_collection": collection}
     if engine == RagEngine.Custom:
-        model = "michaelfeil/bge-small-en-v1.5"
         chat = input[input_field]
-        embeddings = [embedding(input=[r[db_field]], model=model) for r in db]
-        q = embedding(input=[chat], model=model)
+        embeddings = [embedding(input=[r[db_field]]) for r in db]
+        q = embedding(input=[chat])
 
         def cosine_similarity(a, b):
             return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))