change MODEL_NAME to OLLAMA_MODEL_NAME to prevent runpod from blocking deploy

.runpod/hub.json

@@ -12,7 +12,7 @@
     "presets": [],
     "env": [
       {
-        "key": "MODEL_NAME",
+        "key": "OLLAMA_MODEL_NAME",
         "input": {
           "name": "Model Name",
           "type": "string",

.runpod/tests.json

@@ -13,7 +13,7 @@
     "gpuCount": 1,
     "env": [
       {
-        "key": "MODEL_NAME",
+        "key": "OLLAMA_MODEL_NAME",
         "value": "phi3"
       }
     ],

README.md

@@ -2,16 +2,16 @@
 
 ## How to use
 
-Start a runpod serverless with the docker container ``svenbrnn/runpod-ollama:latest``. Set ``MODEL_NAME`` environment to a model from ollama.com to automatically download a model.
+Start a runpod serverless with the docker container ``svenbrnn/runpod-ollama:latest``. Set ``OLLAMA_MODEL_NAME`` environment to a model from ollama.com to automatically download a model.
 A mounted volume will be automatically used.
 
 [![RunPod](https://api.runpod.io/badge/SvenBrnn/runpod-worker-ollama)](https://www.runpod.io/console/hub/SvenBrnn/runpod-worker-ollama)
 
 ## Environment variables
 
-| Variable Name | Description                              | Default Value       |
-|---------------|------------------------------------------|---------------------|
-| `MODEL_NAME`  | The name of the model to download        | NULL                |
+| Variable Name       | Description                              | Default Value       |
+|---------------------|------------------------------------------|---------------------|
+| `OLLAMA_MODEL_NAME` | The name of the model to download        | NULL                |
 
 ## Test requests for runpod.io console
 

src/engine.py

@@ -18,8 +18,8 @@ class OllamaEngine:
         print ("OllamaEngine initialized")
 
     async def generate(self, job_input):
-        # Get model from MODEL_NAME defauting to llama3.2:1b
-        model = os.getenv("MODEL_NAME", "llama3.2:1b")
+        # Get model from OLLAMA_MODEL_NAME defauting to llama3.2:1b
+        model = os.getenv("OLLAMA_MODEL_NAME", "llama3.2:1b")
 
         # Depending if prompt is a string or a list, we need to handle it differently and send it to the OpenAI API
         if isinstance(job_input.llm_input, str):