Skip to content

armgit / dspy

Public Imported from GitHub
0

DSPy: The framework for programming—not prompting—language models

Python

Update minimal-example.md

Omar Khattab19690aea6b1a
diff --git a/docs/quick-start/minimal-example.md b/docs/quick-start/minimal-example.md
index 48894a548..113a0157a 100644
--- a/docs/quick-start/minimal-example.md
+++ b/docs/quick-start/minimal-example.md
@@ -14,21 +14,20 @@ Before we delve into the example, let's ensure our environment is properly confi
```python
import dspy
-from dspy.evaluate import Evaluate
from dspy.datasets.gsm8k import GSM8K, gsm8k_metric
-from dspy.teleprompt import BootstrapFewShotWithRandomSearch
-gms8k = GSM8K()
+# Set up the LM
turbo = dspy.OpenAI(model='gpt-3.5-turbo-instruct', max_tokens=250)
+dspy.settings.configure(lm=turbo)
+# Load math questions from the GSM8K dataset
+gms8k = GSM8K()
trainset, devset = gms8k.train, gms8k.dev
-
-dspy.settings.configure(lm=turbo)
```
## Define the Module
-With our environment set up, let's define a custom module that utilizes the `ChainOfThought` module to perform step-by-step logical reasoning to generate answers to questions:
+With our environment set up, let's define a custom module that utilizes the `ChainOfThought` module to perform step-by-step reasoning to generate answers:
```python
class CoT(dspy.Module):
@@ -42,16 +41,33 @@ class CoT(dspy.Module):
## Compile and Evaluate the Model
-With our custom module in place, let's move on to optimizing the program by compiling the module using the `BootstrapFewShotWithRandomSearch` teleprompter and evaluating its performance on the dev dataset:
+With our custom module in place, let's move on to optimizing the program by compiling the module using the `BootstrapFewShotWithRandomSearch` teleprompter:
```python
-evaluate = Evaluate(devset=devset[:], metric=gsm8k_metric, num_threads=4, display_progress=True, display_table=0)
+from dspy.teleprompt import BootstrapFewShotWithRandomSearch
+# Set up the optimizer: we want to "bootstrap" (i.e., self-generate) 8-shot examples of our CoT programs.
+# The optimizer will repeat this 10 times (plus some initial attempts) before selecting its best attempt on the devset.
config = dict(max_bootstrapped_demos=8, max_labeled_demos=8, num_candidate_programs=10, num_threads=NUM_THREADS)
+
+# Optimize! Use the `gms8k_metric` here. In general, the metric is going to tell the optimizer how well it's doing.
teleprompter = BootstrapFewShotWithRandomSearch(metric=gsm8k_metric, **config)
-cot_bs = teleprompter.compile(CoT(), trainset=trainset, valset=devset)
+optimized_cot = teleprompter.compile(CoT(), trainset=trainset, valset=devset)
+```
+
+## Evaluate
+
+Now that we have a compiled (optimized) DSPy program, let's move to evaluating its performance on the dev dataset.
+
+
+```python
+from dspy.evaluate import Evaluate
+
+# Set up the evaluator, which can be used multiple times.
+evaluate = Evaluate(devset=devset, metric=gsm8k_metric, num_threads=4, display_progress=True, display_table=0)
-evaluate(cot_bs, devset=devset[:])
+# Evaluate our `optimized_cot` program.
+evaluate(optimized_cot)
```
## Inspect the Model's History