Local II

Author: KillianLucas

interpreter/terminal_interface/profiles/defaults/codestral-os.py

@@ -3,6 +3,42 @@
 
 from interpreter import interpreter
 
+interpreter.llm.model = "ollama/codestral"
+interpreter.llm.max_tokens = 1000
+interpreter.llm.context_window = 7000
+
+model_name = interpreter.llm.model.replace("ollama/", "")
+try:
+    # List out all downloaded ollama models. Will fail if ollama isn't installed
+    result = subprocess.run(
+        ["ollama", "list"], capture_output=True, text=True, check=True
+    )
+except Exception as e:
+    print(str(e))
+    interpreter.display_message(
+        f"> Ollama not found\n\nPlease download Ollama from [ollama.com](https://ollama.com/) to use `codestral`.\n"
+    )
+    exit()
+
+lines = result.stdout.split("\n")
+names = [
+    line.split()[0].replace(":latest", "") for line in lines[1:] if line.strip()
+]  # Extract names, trim out ":latest", skip header
+
+if model_name not in names:
+    interpreter.display_message(f"\nDownloading {model_name}...\n")
+    subprocess.run(["ollama", "pull", model_name], check=True)
+
+# Send a ping, which will actually load the model
+interpreter.display_message("\n*Loading model...*\n")
+
+old_max_tokens = interpreter.llm.max_tokens
+interpreter.llm.max_tokens = 1
+interpreter.computer.ai.chat("ping")
+interpreter.llm.max_tokens = old_max_tokens
+
+interpreter.display_message("> Model set to `codestral`")
+
 # Check if required packages are installed
 
 # THERE IS AN INCONSISTENCY HERE.

interpreter/terminal_interface/profiles/defaults/codestral.py

@@ -3,19 +3,38 @@
 from interpreter import interpreter
 
 interpreter.llm.model = "ollama/codestral"
+interpreter.llm.max_tokens = 1000
+interpreter.llm.context_window = 7000
+
+model_name = interpreter.llm.model.replace("ollama/", "")
+try:
+    # List out all downloaded ollama models. Will fail if ollama isn't installed
+    result = subprocess.run(
+        ["ollama", "list"], capture_output=True, text=True, check=True
+    )
+except Exception as e:
+    print(str(e))
+    interpreter.display_message(
+        f"> Ollama not found\n\nPlease download Ollama from [ollama.com](https://ollama.com/) to use `codestral`.\n"
+    )
+    exit()
+
+lines = result.stdout.split("\n")
+names = [
+    line.split()[0].replace(":latest", "") for line in lines[1:] if line.strip()
+]  # Extract names, trim out ":latest", skip header
+
+if model_name not in names:
+    interpreter.display_message(f"\nDownloading {model_name}...\n")
+    subprocess.run(["ollama", "pull", model_name], check=True)
 
 # Send a ping, which will actually load the model
-interpreter.display_message("\nLoading model...")
+interpreter.display_message("\n*Loading model...*\n")
 
 old_max_tokens = interpreter.llm.max_tokens
-old_context_window = interpreter.llm.context_window
 interpreter.llm.max_tokens = 1
-interpreter.llm.context_window = 100
-
 interpreter.computer.ai.chat("ping")
-
 interpreter.llm.max_tokens = old_max_tokens
-interpreter.llm.context_window = old_context_window
 
 interpreter.display_message("> Model set to `codestral`")
 

interpreter/terminal_interface/profiles/defaults/llama3.py

@@ -151,25 +151,40 @@
 Now, your turn:"""
 
 
+interpreter.llm.model = "ollama/llama3"
+interpreter.llm.max_tokens = 1000
+interpreter.llm.context_window = 7000
+
+# The below should be pushed into interpreter.llm.load()?
+
+model_name = interpreter.llm.model.replace("ollama/", "")
 try:
     # List out all downloaded ollama models. Will fail if ollama isn't installed
     result = subprocess.run(
         ["ollama", "list"], capture_output=True, text=True, check=True
     )
-    lines = result.stdout.split("\n")
-    names = [
-        line.split()[0].replace(":latest", "") for line in lines[1:] if line.strip()
-    ]  # Extract names, trim out ":latest", skip header
-
-    if "llama3" not in names:
-        interpreter.display_message(f"\nDownloading llama3...\n")
-        subprocess.run(["ollama", "pull", "llama3"], check=True)
-
-    # Set the model to codestral
-    interpreter.llm.model = f"ollama/llama3"
-    interpreter.display_message(f"> Model set to `llama3`")
-except:
+except Exception as e:
+    print(str(e))
     interpreter.display_message(
-        f"> Ollama not found\n\nPlease download Ollama from [ollama.com](https://ollama.com/) to use `codestral`.\n"
+        f"> Ollama not found\n\nPlease download Ollama from [ollama.com](https://ollama.com/) to use `llama3`.\n"
     )
     exit()
+
+lines = result.stdout.split("\n")
+names = [
+    line.split()[0].replace(":latest", "") for line in lines[1:] if line.strip()
+]  # Extract names, trim out ":latest", skip header
+
+if model_name not in names:
+    interpreter.display_message(f"\nDownloading {model_name}...\n")
+    subprocess.run(["ollama", "pull", model_name], check=True)
+
+# Send a ping, which will actually load the model
+interpreter.display_message("\n*Loading model...*\n")
+
+old_max_tokens = interpreter.llm.max_tokens
+interpreter.llm.max_tokens = 1
+interpreter.computer.ai.chat("ping")
+interpreter.llm.max_tokens = old_max_tokens
+
+interpreter.display_message("> Model set to `llama3`")

interpreter/terminal_interface/validate_llm_settings.py

@@ -1,5 +1,6 @@
 """
 I do not like this and I want to get rid of it lol. Like, what is it doing..?
+I guess it's setting up the model. So maybe this should be like, interpreter.llm.load() soon
 """
 
 import os
@@ -92,7 +93,7 @@ def validate_llm_settings(interpreter):
                 except Exception as e:
                     print(str(e))
                     interpreter.display_message(
-                        f"> Ollama not found\n\nPlease download Ollama from [ollama.com](https://ollama.com/) to use `codestral`.\n"
+                        f"> Ollama not found\n\nPlease download Ollama from [ollama.com](https://ollama.com/) to use `{model_name}`.\n"
                     )
                     exit()
 
@@ -108,15 +109,13 @@ def validate_llm_settings(interpreter):
                     subprocess.run(["ollama", "pull", model_name], check=True)
 
                 # Send a ping, which will actually load the model
-                print("Loading model...")
+                interpreter.display_message("\n*Loading model...*\n")
 
                 old_max_tokens = interpreter.llm.max_tokens
                 interpreter.llm.max_tokens = 1
                 interpreter.computer.ai.chat("ping")
                 interpreter.llm.max_tokens = old_max_tokens
 
-                print("Model loaded.")
-
                 # interpreter.display_message(f"> Model set to `{model_name}`")
 
             # This is a model we don't have checks for yet.