Finalize model selection demo

README.md

@@ -71,6 +71,7 @@ This is an early work-in-progress. Follow [me on twitter](https://x.com/klntsky)
 - [ ] Runtime system
   - [x] Support variable definition at runtime
   - [x] dynamic model switching (via `MODEL` variable - [example](./python/examples/model-change.metaprompt))
+  - [ ] Multiple chat instances and ability to switch between them, to distribute data between chat contexts. E.g. `[$chat1: the object is the moon][$chat2: the object is the sun][$chat1: what is the object?]`
   - [ ] exceptions
     - [ ] throwing exceptions
     - [ ] recovering from exceptions

docs/index.md

@@ -30,9 +30,9 @@ MetaPrompt's basic use case is substituting parameter values instead of variable
 Write me a poem about [:subject] in the style of [:style]
 ```
 
-## Meta-prompting
+## Prompt rewriting
 
-Meta-prompting is a technique of asking an LLM to create/modify/expand an LLM prompt.
+Prompt rewriting is a technique of asking an LLM to create/modify/expand an LLM prompt.
 
 - Dynamically crafting task-specific prompts based on a set of high level principles
 - Modifying prompts to increase accuracy

python/examples/choose-model.metaprompt

@@ -0,0 +1,10 @@
+[$ You must choose the best LLM for a given prompt, considering all the options listed below:
+
+- `gpt-3.5-turbo` - Best for tasks that don't require reasoning, like data processing or generation.
+- `gpt-4o` - Will offer a very high level of intelligence and strong performance. Suitable for programming tasks that require reasoning about software architecture.
+- `gpt-4o-mini` - Suitable for programming tasks that involve straightforward implementations, like utility functions.
+
+Output format: output ONLY the model ID without qoutes.
+
+The prompt: [:prompt]
+]

python/examples/model-selection-demo.metaprompt

@@ -0,0 +1,3 @@
+[:MODEL=[:use ./choose-model :prompt=[:prompt]]]
+Selected model: [:MODEL]
+[$[:prompt]]

python/src/runtime.py

@@ -12,7 +12,7 @@ def __init__(self, config, env):
         self.cwd = os.getcwd()
 
     def get_current_model(self):
-        return self.env.get("MODEL")
+        return self.env.get("MODEL").strip()
 
     def set_variable(self, var_name, value):
         self.env.set(var_name, value)