Merge pull request #359 from aitomatic/release

Release 0.24.9.30

examples/FinanceBench/visualization.py

@@ -0,0 +1,101 @@
+import numpy as np
+import matplotlib.pyplot as plt
+
+
+# Define the categories
+categories = ['0-RETRIEVE', '1-COMPARE', '2-CALC-CHANGE',
+              '3-CALC-COMPLEX', '4-CALC-AND-JUDGE', '5-EXPLAIN-FACTORS', '6-OTHER-ADVANCED']
+
+
+# Create angles for the plot
+N = len(categories)
+angles = [n / float(N) * 2 * np.pi for n in range(N)]
+# Calculate radii for each difficulty level with increased dynamic range
+max_radius = 10
+radii = [max_radius * (1 + i * 0.8) for i in range(N)]
+# Define percentage zones and colors for background
+percentages = [0.25, 0.5, 0.75, 1.0]
+colors = ['#FFF3E0', '#FFE0B2', '#FFCC80', '#FFB74D']
+
+
+# Accuracy Data
+accuracy_llamaindex_rag = [71, 83, 78, 30, 14, 70, 43]
+accuracy_langchain_react = [85, 90, 69, 88, 60, 70, 37]
+accuracy_openai_assistant = [49, 46, 36, 40, 14, 50, 46]
+accuracy_openssa_dana = [95, 90, 93, 100, 94, 100, 89]
+
+# Consistency Data
+consistency_llamaindex_rag = [96, 90, 100, 87, 88, 60, 100]
+consistency_langchain_react = [86, 91, 78, 84, 68, 60, 66]
+consistency_openai_assistant = [76, 65, 87, 73, 72, 100, 77]
+consistency_openssa_dana = [96, 97, 87, 100, 92, 100, 94]
+
+
+# Function to plot the data
+def plot_data(ax, data, color, label=None):
+    values = data
+    r = [radius * value / 100 for radius, value in zip(radii, values)]
+    ax.plot(angles + [angles[0]], r + [r[0]], 'o-', linewidth=2, label=label, color=color)
+    ax.fill(angles + [angles[0]], r + [r[0]], color=color, alpha=0.25)
+
+
+# Create the figure and two subplots
+fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(18, 10), subplot_kw={'projection': 'polar'})
+
+
+# Function to set up the polar plot (background, grid, labels)
+def setup_polar_plot(ax, title):
+    # Remove grid and spines
+    ax.grid(False)
+    ax.spines['polar'].set_visible(False)
+    # Plot radial lines (terminating at 100%)
+    for angle, radius in zip(angles, radii):
+        ax.plot([angle, angle], [0, radius], color='gray', linestyle='-', linewidth=0.5, alpha=0.5)
+    # Plot percentage lines and fill background
+    for i, percentage in enumerate(percentages):
+        r = [radius * percentage for radius in radii]
+        theta = np.linspace(0, angles[-1], 1000)
+        spiral_r = np.interp(theta, angles, r)
+        ax.plot(theta, spiral_r, color='gray', linestyle='--', linewidth=0.5, alpha=0.7)
+        if i > 0:
+            ax.fill(angles + [angles[0]], r + [r[0]], colors[i], alpha=0.1)
+    # Add full category labels without rotation
+    for i, category in enumerate(categories):
+        angle = angles[i]
+        r = radii[i] * 1.15
+        ha = 'left' if -np.pi / 2 <= angle < np.pi / 2 else 'right'
+        ax.text(angle, r, category, ha=ha, va='center', fontsize=10, fontweight='bold')
+    # Remove radial labels and ticks
+    ax.set_yticks([])
+    ax.set_yticklabels([])
+    ax.set_xticks([])
+    ax.set_xticklabels([])
+    # Set the ylim to ensure all labels are visible
+    ax.set_ylim(0, max(radii) * 1.3)
+    # Set title for the chart
+    ax.set_title(title, y=0.75, fontsize=14, fontweight='bold')
+
+
+# Set up both subplots
+setup_polar_plot(ax1, "AGENT ACCURACY")
+setup_polar_plot(ax2, "AGENT CONSISTENCY")
+
+# Plot accuracy data on the first subplot with labels for the legend
+plot_data(ax1, accuracy_llamaindex_rag, 'red', 'LlamaIndex RAG')
+plot_data(ax1, accuracy_langchain_react, 'yellow', 'LangChain ReAct')
+plot_data(ax1, accuracy_openai_assistant, 'blue', 'OpenAI Assistant')
+plot_data(ax1, accuracy_openssa_dana, 'green', 'OpenSSA DANA-NK-NP')
+
+# Plot consistency data on the second subplot without labels to avoid duplicate legend
+plot_data(ax2, consistency_llamaindex_rag, 'red')
+plot_data(ax2, consistency_langchain_react, 'yellow')
+plot_data(ax2, consistency_openai_assistant, 'blue')
+plot_data(ax2, consistency_openssa_dana, 'green')
+
+# Adjust space between subplots
+plt.subplots_adjust(wspace=-0.4)
+
+# Add a single legend closer to the plots
+fig.legend(loc='lower center', bbox_to_anchor=(0.5, 0), fontsize=12, ncol=4)
+plt.tight_layout()
+plt.savefig('agent_comparison.pdf', bbox_inches='tight')

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "openssa"
-version = "0.24.9.25"
+version = "0.24.9.30"
 
 authors = [
   "Aitomatic, Inc. <engineering@aitomatic.com>",
@@ -12,9 +12,9 @@ maintainers = [
   "Aitomatic, Inc. <engineering@aitomatic.com>",
   "Vinh Luong <vinh@aitomatic.com>",
   "Sang Dinh <sang@aitomatic.com>",
-  "Quynh Le <quynh@aitomatic.com>",
   "Shruti Raghavan <shruti@aitomatic.com>",
   "William Nguyen <william@aitomatic.com>",
+  "Quynh Le <quynh@aitomatic.com>",
   "Zooey Nguyen <zooey@aitomatic.com>",
 ]
 
@@ -85,6 +85,7 @@ adlfs = ">=2024.7"  # Azure Data Lake
 gcsfs = ">=2024.9"  # Google Cloud Storage
 s3fs = ">=2024.9"  # S3
 # misc / other
+certifi = ">2023.7.22"
 google-api-python-client = ">=2.146"
 googlesearch-python = ">=1.2"
 httpx = ">=0.27"