simonbb 0.0.7__tar.gz → 0.0.9__tar.gz

{simonbb-0.0.7 → simonbb-0.0.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: simonbb
-Version: 0.0.7
+Version: 0.0.9
 Summary: A small package to facilitate teaching AI courses at business schools
 Project-URL: Homepage, https://simonbb.com
 Author-email: Huaxia Rui <huaxia.rui@simon.rochester.edu>
@@ -9,24 +9,17 @@ License-File: LICENSE
 Classifier: Operating System :: OS Independent
 Classifier: Programming Language :: Python :: 3
 Requires-Python: >=3.9
-Requires-Dist: gymnasium~=1.3.0
-Requires-Dist: langchain-chroma~=1.1.0
-Requires-Dist: langchain-community~=0.4.1
-Requires-Dist: langchain-core~=1.2.23
-Requires-Dist: langchain-huggingface~=1.2.1
-Requires-Dist: langchain~=1.2.13
-Requires-Dist: langgraph~=1.1.3
-Requires-Dist: matplotlib~=3.10.8
-Requires-Dist: numpy~=2.4.0
-Requires-Dist: opencv-python~=5.0.0.93
-Requires-Dist: pytorch-ignite~=0.5.4
-Requires-Dist: pyyaml~=6.0.3
-Requires-Dist: ray~=2.56.0
-Requires-Dist: sentence-transformers~=5.3.0
-Requires-Dist: stable-baselines3~=2.9.0
-Requires-Dist: torchvision~=0.24.1
-Requires-Dist: torch~=2.9.1
-Requires-Dist: transformers~=4.57.3
+Requires-Dist: gymnasium>=1.3.0
+Requires-Dist: matplotlib>=3.10.8
+Requires-Dist: numpy>=2.4.0
+Requires-Dist: opencv-python>=5.0.0.93
+Requires-Dist: pytorch-ignite>=0.5.4
+Requires-Dist: pyyaml>=6.0.3
+Requires-Dist: ray>=2.56.0
+Requires-Dist: stable-baselines3>=2.9.0
+Requires-Dist: torch>=2.9.1
+Requires-Dist: torchvision>=0.24.1
+Requires-Dist: transformers>=4.57.3
 Description-Content-Type: text/markdown
 
 # Background

{simonbb-0.0.7 → simonbb-0.0.9}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "simonbb"
-version = "0.0.7"
+version = "0.0.9"
 authors = [
 	{ name="Huaxia Rui", email="huaxia.rui@simon.rochester.edu" },
 ]
@@ -18,24 +18,17 @@ classifiers = [
 license = "MIT"
 license-files = ["LICEN[CS]E*"]
 dependencies = [
-	"matplotlib~=3.10.8",
-	"torch~=2.9.1",
-	"torchvision~=0.24.1",
-	"numpy~=2.4.0",
-	"transformers~=4.57.3",
-	"langchain-core~=1.2.23",
-	"langchain~=1.2.13",
-	"langgraph~=1.1.3",
-	"langchain-community~=0.4.1",
-	"langchain-chroma~=1.1.0",
-	"langchain-huggingface~=1.2.1",
-	"sentence-transformers~=5.3.0",
-	"gymnasium~=1.3.0",
-	"pytorch-ignite~=0.5.4",
-	"stable_baselines3~=2.9.0",
-	"ray~=2.56.0",
-	"opencv-python~=5.0.0.93",
-	"pyyaml~=6.0.3"
+	"matplotlib >= 3.10.8",
+	"torch >= 2.9.1",
+	"torchvision >=0.24.1",
+	"numpy >=2.4.0",
+	"transformers >= 4.57.3",
+	"gymnasium >= 1.3.0",
+	"pytorch-ignite >= 0.5.4",
+	"stable_baselines3 >= 2.9.0",
+	"ray >= 2.56.0",
+	"opencv-python >= 5.0.0.93",
+	"pyyaml >= 6.0.3"
 ]
 [project.urls]
 Homepage = "https://simonbb.com"

{simonbb-0.0.7 → simonbb-0.0.9}/src/simonbb/cis433/experience.py

@@ -31,9 +31,7 @@ class ExperienceSource:
     """
     Item = tt.Tuple[Experience, ...]
 
-    def __init__(self, env: gym.Env | tt.Collection[gym.Env], agent: BaseAgent,
-                 steps_count: int = 2, steps_delta: int = 1,
-                 env_seed: tt.Optional[int] = None):
+    def __init__(self, env: gym.Env | tt.Collection[gym.Env], agent: BaseAgent, steps_count: int = 2, steps_delta: int = 1, env_seed: tt.Optional[int] = None):
         """
         Create simple experience source
         :param env: environment or list of environments to be used
@@ -46,7 +44,7 @@ class ExperienceSource:
         if isinstance(env, (list, tuple)):
             self.pool = env
             # do the check for the multiple copies passed
-            ids = set(id(e) for e in env)
+            ids = set(id(e) for e in env) # id() returns a unique, constant integer identifier for a specified object during its lifetime
             if len(ids) < len(env):
                 raise ValueError("You passed single environment instance multiple times")
         else:

simonbb-0.0.9/src/simonbb/cis433/math500.py

@@ -0,0 +1,105 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+import re
+import time
+import json
+from pathlib import Path
+from ..utils import grade_math_answer, eta_progress_message
+
+RE_NUMBER = re.compile( r"-?(?:\d+/\d+|\d+(?:\.\d+)?(?:[eE][+-]?\d+)?)" )
+def get_last_boxed(text):
+    # Find the last occurrence of "\boxed"
+    boxed_start_idx = text.rfind(r"\boxed")
+    if boxed_start_idx == -1:
+        return None
+    # Get position after "\boxed"
+    current_idx = boxed_start_idx + len(r"\boxed")
+    # Skip any whitespace after "\boxed"
+    while current_idx < len(text) and text[current_idx].isspace():
+        current_idx += 1
+    # Expect an opening brace "{"
+    if current_idx >= len(text) or text[current_idx] != "{":
+        return None
+    # Parse the braces with nesting
+    current_idx += 1
+    brace_depth = 1
+    content_start_idx = current_idx
+    while current_idx < len(text) and brace_depth > 0:
+        char = text[current_idx]
+        if char == "{":
+            brace_depth += 1
+        elif char == "}":
+            brace_depth -= 1
+        current_idx += 1
+    # Account for unbalanced braces
+    if brace_depth != 0:
+        return None
+    # Extract content inside the outermost braces
+    return text[content_start_idx:current_idx-1]
+
+def extract_final_candidate(text, fallback="number_then_full"):
+    # Default return value if nothing matches
+    result = ""
+    if text:        
+        boxed = get_last_boxed(text.strip()) # Prefer the last boxed expression if present
+        if boxed:
+            result = boxed.strip().strip("$ ")
+        # If no boxed expression, try fallback
+        elif fallback in ("number_then_full", "number_only"):
+            m = RE_NUMBER.findall(text)
+            if m: # Use last number
+                result = m[-1]
+            elif fallback == "number_then_full":
+                # Else return full text if no number found
+                result = text
+    return result
+
+def render_prompt(prompt):
+    template = (
+        "You are a helpful math assistant.\n"
+        "Answer the question and write the final result on a new line as:\n"
+        "\\boxed{ANSWER}\n\n"
+        f"Question:\n{prompt}\n\nAnswer:"
+    )
+    return template
+
+
+def evaluate_math500_stream( model, tokenizer, device, generate_func, math_data, out_path=None, max_new_tokens=512, verbose=False ):
+    num_examples = len(math_data)
+    num_correct = 0
+    start_time = time.time()
+    with open(Path( out_path ), "w", encoding="utf-8") as f:  # Save results for inspection
+        for i, row in enumerate(math_data, start=1):
+            prompt = render_prompt(row["problem"])    # 1. Apply prompt template
+            gen_text = generate_func( model=model, tokenizer=tokenizer, prompt=prompt, device=device, max_new_tokens=max_new_tokens, verbose=verbose )
+            extracted = extract_final_candidate( gen_text )
+            is_correct = grade_math_answer( extracted, row["answer"] )
+            num_correct += int(is_correct)
+            record = {  # Record to be saved for inspection
+                "index": i,
+                "problem": row["problem"],
+                "gtruth_answer": row["answer"],
+                "generated_text": gen_text,
+                "extracted": extracted,
+                "correct": bool(is_correct),
+            }
+            f.write(json.dumps(record, ensure_ascii=False) + "\n")
+
+            progress_msg = eta_progress_message( processed=i, total=num_examples, start_time=start_time, show_eta=True, label="MATH-500" )
+            print(progress_msg, end="\r", flush=True)
+            if verbose:  # Print responses during the generation process
+                print(
+                    f"\n\n{'='*50}\n{progress_msg}\n"
+                    f"{'='*50}\nExtracted: {extracted}\n"
+                    f"Expected:  {row['answer']}\n"
+                    f"Correct so far: {num_correct}\n{'-'*50}"
+                )
+
+    # Print summary information
+    seconds_elapsed = time.time() - start_time
+    acc = num_correct / num_examples if num_examples else 0.0
+    print(f"\nAccuracy: {acc*100:.1f}% ({num_correct}/{num_examples})")
+    print(f"Total time: {seconds_elapsed/60:.1f} min")
+    print(f"Logs written to: {out_path}")
+    return num_correct, num_examples, acc