promptum 0.0.1__tar.gz → 0.0.3__tar.gz

promptum-0.0.3/CONTRIBUTING.md

@@ -0,0 +1,78 @@
+# Contributing to Promptum
+
+Thank you for your interest in contributing to Promptum! We welcome contributions from the community.
+
+## Getting Started
+
+1. **Fork the repository** to your own GitHub account
+2. **Clone your fork** locally:
+   ```bash
+   git clone https://github.com/YOUR_USERNAME/promptum.git
+   cd promptum
+   ```
+3. **Set up the development environment**:
+   ```bash
+   just sync  # Install/sync dependencies
+   ```
+
+## Making Changes
+
+### Branch Naming
+
+Create a new branch named after the issue number you're working on:
+
+```bash
+git checkout -b 42  # For issue #42
+```
+
+### One PR = One Issue
+
+Each pull request should address exactly one issue. If you want to work on multiple issues, create separate branches and PRs for each.
+
+### Work in Progress
+
+If your PR is not ready for review, add `[WIP]` to the title:
+
+```
+[WIP] #42: Fix retry logic in OpenRouterClient
+```
+
+Remove `[WIP]` when the PR is ready for review.
+
+## Submitting Changes
+
+1. **Run tests and linting** before committing:
+   ```bash
+   just lint       # Lint and auto-fix
+   just typecheck  # Type check
+   just test       # Run tests
+   ```
+
+2. **Commit your changes** with clear, descriptive messages:
+   ```bash
+   git commit -m "#42: Fix retry logic in OpenRouterClient"
+   ```
+
+3. **Push to your fork**:
+   ```bash
+   git push origin 42
+   ```
+
+4. **Create a Pull Request** from your fork to the main repository
+
+5. **Tag the maintainer** (@deyna256) in a comment when your PR is ready for review
+
+## CI Requirements
+
+Pull requests must pass all CI checks before review. The maintainer will not review PRs with failing checks.
+
+CI runs:
+- Linting
+- Type checking
+- Tests
+
+## Questions?
+
+Feel free to ask questions in the issue comments or open a discussion.
+
+Thank you for contributing!

{promptum-0.0.1 → promptum-0.0.3}/Justfile

@@ -23,10 +23,6 @@ cov-html:
     uv run pytest tests/ --cov-report=html
     xdg-open htmlcov/index.html
 
-# Open benchmark HTML report
-report:
-    xdg-open results/report.html
-
 # Clean up generated files and caches
 clean:
     rm -rf .pytest_cache .ruff_cache .coverage htmlcov results/

{promptum-0.0.1 → promptum-0.0.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: promptum
-Version: 0.0.1
+Version: 0.0.3
 Summary: Async LLM benchmarking library with protocol-based extensibility
 Project-URL: Homepage, https://github.com/deyna256/promptum
 Project-URL: Repository, https://github.com/deyna256/promptum
@@ -36,8 +36,6 @@ Classifier: Programming Language :: Python :: 3.13
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Requires-Python: >=3.13
 Requires-Dist: httpx>=0.27.0
-Requires-Dist: jinja2>=3.1.0
-Requires-Dist: pyyaml>=6.0
 Description-Content-Type: text/markdown
 
 # promptum
@@ -48,7 +46,7 @@ Description-Content-Type: text/markdown
 ![Async](https://img.shields.io/badge/Async-First-green?style=for-the-badge)
 ![License: MIT](https://img.shields.io/badge/License-MIT-yellow?style=for-the-badge)
 
-**Benchmark LLMs Like a Pro. In 5 Lines of Code.**
+**Benchmark LLMs Like a Pro.**
 
 Stop writing boilerplate to test LLMs. Start getting results.
 
@@ -58,11 +56,12 @@ Stop writing boilerplate to test LLMs. Start getting results.
 
 ## What's This?
 
-A dead-simple Python library for benchmarking LLM providers. Write tests once, run them across any model, get beautiful reports.
+A dead-simple Python library for benchmarking LLM providers. Write tests once, run them across any model, get structured results.
 
 ```python
 benchmark = Benchmark(provider=client, name="my_test")
 benchmark.add_test(TestCase(
+    name="basic_math",
     prompt="What is 2+2?",
     model="gpt-3.5-turbo",
     validator=Contains("4")
@@ -97,15 +96,12 @@ for attempt in range(max_retries):
         break
     except Exception:
         sleep(2 ** attempt)
-
-# Export results manually
-json.dump(results, open("results.json", "w"))
 ```
 
 **After promptum:**
 ```python
 report = await benchmark.run_async()
-HTMLSerializer().serialize(report)  # Beautiful HTML report
+summary = report.get_summary()  # Metrics captured automatically
 ```
 
 ---
@@ -135,9 +131,9 @@ async def main():
         report = await benchmark.run_async()
         summary = report.get_summary()
 
-        print(f"✓ {summary['passed']}/{summary['total']} tests passed")
-        print(f"⚡ {summary['avg_latency_ms']:.0f}ms average")
-        print(f"💰 ${summary['total_cost_usd']:.6f} total cost")
+        print(f"✓ {summary.passed}/{summary.total} tests passed")
+        print(f"⚡ {summary.avg_latency_ms:.0f}ms average")
+        print(f"💰 ${summary.total_cost_usd:.6f} total cost")
 
 asyncio.run(main())
 ```
@@ -151,14 +147,13 @@ python your_script.py
 
 ## What You Get
 
-✅ **One API for 100+ Models** - OpenRouter support out of the box (OpenAI, Anthropic, Google, etc.)
-✅ **Smart Validation** - ExactMatch, Contains, Regex, JsonSchema, or write your own
-✅ **Automatic Retries** - Exponential/linear backoff with configurable attempts
-✅ **Metrics Tracking** - Latency, tokens, cost - automatically captured
-✅ **Beautiful Reports** - JSON, YAML, or interactive HTML with charts
-✅ **Async by Default** - Run 100 tests in parallel without breaking a sweat
-✅ **Type Safe** - Full type hints, catches errors before runtime
-✅ **Zero Config** - No YAML files, no setup scripts, just Python
+- [x] **100+ Models via OpenRouter** - One client for OpenAI, Anthropic, Google, and more
+- [x] **Smart Validation** - ExactMatch, Contains, Regex, JsonSchema, or write your own
+- [x] **Automatic Retries** - Exponential/linear backoff with configurable attempts
+- [x] **Metrics Tracking** - Latency, tokens, cost - automatically captured
+- [x] **Async by Default** - Run 100 tests in parallel without breaking a sweat
+- [x] **Type Safe** - Full type hints, catches errors before runtime
+- [x] **Zero Config** - No YAML files, no setup scripts, just Python
 
 ---
 
@@ -167,39 +162,43 @@ python your_script.py
 Compare GPT-4 vs Claude on your tasks:
 
 ```python
-from promptum import Benchmark, TestCase, ExactMatch, Contains, Regex
-
-tests = [
-    TestCase(
-        name="json_output",
-        prompt='Output JSON: {"status": "ok"}',
-        model="openai/gpt-4",
-        validator=Regex(r'\{"status":\s*"ok"\}')
-    ),
-    TestCase(
-        name="json_output",
-        prompt='Output JSON: {"status": "ok"}',
-        model="anthropic/claude-3-5-sonnet",
-        validator=Regex(r'\{"status":\s*"ok"\}')
-    ),
-    TestCase(
-        name="creative_writing",
-        prompt="Write a haiku about Python",
-        model="openai/gpt-4",
-        validator=Contains("Python", case_sensitive=False)
-    ),
-]
-
-benchmark.add_tests(tests)
-report = await benchmark.run_async()
+import asyncio
+from promptum import Benchmark, TestCase, Contains, Regex, OpenRouterClient
 
-# Export as HTML
-from promptum import HTMLSerializer
-html = HTMLSerializer().serialize(report)
-open("comparison.html", "w").write(html)
-```
+async def main():
+    async with OpenRouterClient(api_key="your-key") as client:
+        benchmark = Benchmark(provider=client, name="model_comparison")
+
+        benchmark.add_tests([
+            TestCase(
+                name="json_output_gpt4",
+                prompt='Output JSON: {"status": "ok"}',
+                model="openai/gpt-4",
+                validator=Regex(r'\{"status":\s*"ok"\}')
+            ),
+            TestCase(
+                name="json_output_claude",
+                prompt='Output JSON: {"status": "ok"}',
+                model="anthropic/claude-3-5-sonnet",
+                validator=Regex(r'\{"status":\s*"ok"\}')
+            ),
+            TestCase(
+                name="creative_writing",
+                prompt="Write a haiku about Python",
+                model="openai/gpt-4",
+                validator=Contains("Python", case_sensitive=False)
+            ),
+        ])
 
-Open `comparison.html` in your browser - see side-by-side model performance with charts.
+        report = await benchmark.run_async()
+
+        # Side-by-side model comparison
+        for model, model_report in report.group_by(lambda r: r.test_case.model).items():
+            summary = model_report.get_summary()
+            print(f"{model}: {summary.pass_rate:.0%} pass rate, {summary.avg_latency_ms:.0f}ms avg")
+
+asyncio.run(main())
+```
 
 ---
 
@@ -252,7 +251,7 @@ Found a bug? Want a feature? PRs welcome!
 
 ```bash
 # Development setup
-git clone https://github.com/yourusername/promptum.git
+git clone https://github.com/deyna256/promptum.git
 cd promptum
 just sync       # Install dependencies
 just test       # Run tests
@@ -273,7 +272,7 @@ MIT - do whatever you want with it.
 
 <div align="center">
 
-**[⭐ Star on GitHub](https://github.com/yourusername/promptum)** | **[🐛 Report Bug](https://github.com/yourusername/promptum/issues)** | **[💡 Request Feature](https://github.com/yourusername/promptum/issues)**
+**[⭐ Star on GitHub](https://github.com/deyna256/promptum)** | **[🐛 Report Bug](https://github.com/deyna256/promptum/issues)** | **[💡 Request Feature](https://github.com/deyna256/promptum/issues)**
 
 Made for developers who value their time.
 

{promptum-0.0.1 → promptum-0.0.3}/README.md

@@ -6,7 +6,7 @@
 ![Async](https://img.shields.io/badge/Async-First-green?style=for-the-badge)
 ![License: MIT](https://img.shields.io/badge/License-MIT-yellow?style=for-the-badge)
 
-**Benchmark LLMs Like a Pro. In 5 Lines of Code.**
+**Benchmark LLMs Like a Pro.**
 
 Stop writing boilerplate to test LLMs. Start getting results.
 
@@ -16,11 +16,12 @@ Stop writing boilerplate to test LLMs. Start getting results.
 
 ## What's This?
 
-A dead-simple Python library for benchmarking LLM providers. Write tests once, run them across any model, get beautiful reports.
+A dead-simple Python library for benchmarking LLM providers. Write tests once, run them across any model, get structured results.
 
 ```python
 benchmark = Benchmark(provider=client, name="my_test")
 benchmark.add_test(TestCase(
+    name="basic_math",
     prompt="What is 2+2?",
     model="gpt-3.5-turbo",
     validator=Contains("4")
@@ -55,15 +56,12 @@ for attempt in range(max_retries):
         break
     except Exception:
         sleep(2 ** attempt)
-
-# Export results manually
-json.dump(results, open("results.json", "w"))
 ```
 
 **After promptum:**
 ```python
 report = await benchmark.run_async()
-HTMLSerializer().serialize(report)  # Beautiful HTML report
+summary = report.get_summary()  # Metrics captured automatically
 ```
 
 ---
@@ -93,9 +91,9 @@ async def main():
         report = await benchmark.run_async()
         summary = report.get_summary()
 
-        print(f"✓ {summary['passed']}/{summary['total']} tests passed")
-        print(f"⚡ {summary['avg_latency_ms']:.0f}ms average")
-        print(f"💰 ${summary['total_cost_usd']:.6f} total cost")
+        print(f"✓ {summary.passed}/{summary.total} tests passed")
+        print(f"⚡ {summary.avg_latency_ms:.0f}ms average")
+        print(f"💰 ${summary.total_cost_usd:.6f} total cost")
 
 asyncio.run(main())
 ```
@@ -109,14 +107,13 @@ python your_script.py
 
 ## What You Get
 
-✅ **One API for 100+ Models** - OpenRouter support out of the box (OpenAI, Anthropic, Google, etc.)
-✅ **Smart Validation** - ExactMatch, Contains, Regex, JsonSchema, or write your own
-✅ **Automatic Retries** - Exponential/linear backoff with configurable attempts
-✅ **Metrics Tracking** - Latency, tokens, cost - automatically captured
-✅ **Beautiful Reports** - JSON, YAML, or interactive HTML with charts
-✅ **Async by Default** - Run 100 tests in parallel without breaking a sweat
-✅ **Type Safe** - Full type hints, catches errors before runtime
-✅ **Zero Config** - No YAML files, no setup scripts, just Python
+- [x] **100+ Models via OpenRouter** - One client for OpenAI, Anthropic, Google, and more
+- [x] **Smart Validation** - ExactMatch, Contains, Regex, JsonSchema, or write your own
+- [x] **Automatic Retries** - Exponential/linear backoff with configurable attempts
+- [x] **Metrics Tracking** - Latency, tokens, cost - automatically captured
+- [x] **Async by Default** - Run 100 tests in parallel without breaking a sweat
+- [x] **Type Safe** - Full type hints, catches errors before runtime
+- [x] **Zero Config** - No YAML files, no setup scripts, just Python
 
 ---
 
@@ -125,39 +122,43 @@ python your_script.py
 Compare GPT-4 vs Claude on your tasks:
 
 ```python
-from promptum import Benchmark, TestCase, ExactMatch, Contains, Regex
-
-tests = [
-    TestCase(
-        name="json_output",
-        prompt='Output JSON: {"status": "ok"}',
-        model="openai/gpt-4",
-        validator=Regex(r'\{"status":\s*"ok"\}')
-    ),
-    TestCase(
-        name="json_output",
-        prompt='Output JSON: {"status": "ok"}',
-        model="anthropic/claude-3-5-sonnet",
-        validator=Regex(r'\{"status":\s*"ok"\}')
-    ),
-    TestCase(
-        name="creative_writing",
-        prompt="Write a haiku about Python",
-        model="openai/gpt-4",
-        validator=Contains("Python", case_sensitive=False)
-    ),
-]
-
-benchmark.add_tests(tests)
-report = await benchmark.run_async()
+import asyncio
+from promptum import Benchmark, TestCase, Contains, Regex, OpenRouterClient
 
-# Export as HTML
-from promptum import HTMLSerializer
-html = HTMLSerializer().serialize(report)
-open("comparison.html", "w").write(html)
-```
+async def main():
+    async with OpenRouterClient(api_key="your-key") as client:
+        benchmark = Benchmark(provider=client, name="model_comparison")
+
+        benchmark.add_tests([
+            TestCase(
+                name="json_output_gpt4",
+                prompt='Output JSON: {"status": "ok"}',
+                model="openai/gpt-4",
+                validator=Regex(r'\{"status":\s*"ok"\}')
+            ),
+            TestCase(
+                name="json_output_claude",
+                prompt='Output JSON: {"status": "ok"}',
+                model="anthropic/claude-3-5-sonnet",
+                validator=Regex(r'\{"status":\s*"ok"\}')
+            ),
+            TestCase(
+                name="creative_writing",
+                prompt="Write a haiku about Python",
+                model="openai/gpt-4",
+                validator=Contains("Python", case_sensitive=False)
+            ),
+        ])
 
-Open `comparison.html` in your browser - see side-by-side model performance with charts.
+        report = await benchmark.run_async()
+
+        # Side-by-side model comparison
+        for model, model_report in report.group_by(lambda r: r.test_case.model).items():
+            summary = model_report.get_summary()
+            print(f"{model}: {summary.pass_rate:.0%} pass rate, {summary.avg_latency_ms:.0f}ms avg")
+
+asyncio.run(main())
+```
 
 ---
 
@@ -210,7 +211,7 @@ Found a bug? Want a feature? PRs welcome!
 
 ```bash
 # Development setup
-git clone https://github.com/yourusername/promptum.git
+git clone https://github.com/deyna256/promptum.git
 cd promptum
 just sync       # Install dependencies
 just test       # Run tests
@@ -231,7 +232,7 @@ MIT - do whatever you want with it.
 
 <div align="center">
 
-**[⭐ Star on GitHub](https://github.com/yourusername/promptum)** | **[🐛 Report Bug](https://github.com/yourusername/promptum/issues)** | **[💡 Request Feature](https://github.com/yourusername/promptum/issues)**
+**[⭐ Star on GitHub](https://github.com/deyna256/promptum)** | **[🐛 Report Bug](https://github.com/deyna256/promptum/issues)** | **[💡 Request Feature](https://github.com/deyna256/promptum/issues)**
 
 Made for developers who value their time.
 

{promptum-0.0.1 → promptum-0.0.3}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "promptum"
-version = "0.0.1"
+version = "0.0.3"
 description = "Async LLM benchmarking library with protocol-based extensibility"
 readme = "README.md"
 requires-python = ">=3.13"
@@ -18,8 +18,6 @@ classifiers = [
 ]
 dependencies = [
     "httpx>=0.27.0",
-    "pyyaml>=6.0",
-    "jinja2>=3.1.0",
 ]
 
 [project.urls]

promptum-0.0.3/src/promptum/__init__.py

@@ -0,0 +1,30 @@
+from promptum.benchmark import Benchmark, Report, Runner, Summary, TestCase, TestResult
+from promptum.providers import LLMProvider, Metrics, OpenRouterClient, RetryConfig, RetryStrategy
+from promptum.validation import (
+    Contains,
+    ExactMatch,
+    JsonSchema,
+    Regex,
+    Validator,
+)
+
+__version__ = "0.0.3"
+
+__all__ = [
+    "TestCase",
+    "TestResult",
+    "Summary",
+    "Metrics",
+    "RetryConfig",
+    "RetryStrategy",
+    "Validator",
+    "ExactMatch",
+    "Contains",
+    "Regex",
+    "JsonSchema",
+    "LLMProvider",
+    "OpenRouterClient",
+    "Runner",
+    "Benchmark",
+    "Report",
+]

promptum-0.0.3/src/promptum/benchmark/__init__.py

@@ -0,0 +1,8 @@
+from promptum.benchmark.benchmark import Benchmark
+from promptum.benchmark.report import Report
+from promptum.benchmark.result import TestResult
+from promptum.benchmark.runner import Runner
+from promptum.benchmark.summary import Summary
+from promptum.benchmark.test_case import TestCase
+
+__all__ = ["Benchmark", "Report", "Runner", "Summary", "TestCase", "TestResult"]

{promptum-0.0.1 → promptum-0.0.3}/src/promptum/benchmark/benchmark.py

@@ -1,11 +1,10 @@
 import asyncio
 from collections.abc import Callable, Sequence
-from typing import Any
 
 from promptum.benchmark.report import Report
-from promptum.core.result import TestResult
-from promptum.core.test_case import TestCase
-from promptum.execution.runner import Runner
+from promptum.benchmark.result import TestResult
+from promptum.benchmark.runner import Runner
+from promptum.benchmark.test_case import TestCase
 from promptum.providers.protocol import LLMProvider
 
 
@@ -29,12 +28,12 @@ class Benchmark:
     def add_tests(self, test_cases: Sequence[TestCase]) -> None:
         self._test_cases.extend(test_cases)
 
-    def run(self, metadata: dict[str, Any] | None = None) -> Report:
-        return asyncio.run(self.run_async(metadata))
+    def run(self) -> Report:
+        return asyncio.run(self.run_async())
 
-    async def run_async(self, metadata: dict[str, Any] | None = None) -> Report:
+    async def run_async(self) -> Report:
         if not self._test_cases:
-            return Report(results=[], metadata=metadata or {})
+            return Report(results=[])
 
         runner = Runner(
             provider=self.provider,
@@ -44,7 +43,4 @@ class Benchmark:
 
         results = await runner.run(self._test_cases)
 
-        return Report(
-            results=results,
-            metadata=metadata or {},
-        )
+        return Report(results=results)

{promptum-0.0.1 → promptum-0.0.3}/src/promptum/benchmark/report.py

@@ -1,16 +1,15 @@
 from collections.abc import Callable, Sequence
 from dataclasses import dataclass
-from typing import Any
 
-from promptum.core.result import TestResult
+from promptum.benchmark.result import TestResult
+from promptum.benchmark.summary import Summary
 
 
 @dataclass(frozen=True, slots=True)
 class Report:
     results: Sequence[TestResult]
-    metadata: dict[str, Any]
 
-    def get_summary(self) -> dict[str, Any]:
+    def get_summary(self) -> Summary:
         total = len(self.results)
         passed = sum(1 for r in self.results if r.passed)
 
@@ -18,18 +17,17 @@ class Report:
         total_cost = sum(r.metrics.cost_usd or 0 for r in self.results if r.metrics)
         total_tokens = sum(r.metrics.total_tokens or 0 for r in self.results if r.metrics)
 
-        return {
-            "total": total,
-            "passed": passed,
-            "failed": total - passed,
-            "pass_rate": passed / total if total > 0 else 0,
-            "avg_latency_ms": sum(latencies) / len(latencies) if latencies else 0,
-            "p50_latency_ms": self._percentile(latencies, 0.5) if latencies else 0,
-            "p95_latency_ms": self._percentile(latencies, 0.95) if latencies else 0,
-            "p99_latency_ms": self._percentile(latencies, 0.99) if latencies else 0,
-            "total_cost_usd": total_cost,
-            "total_tokens": total_tokens,
-        }
+        return Summary(
+            total=total,
+            passed=passed,
+            failed=total - passed,
+            pass_rate=passed / total if total > 0 else 0,
+            avg_latency_ms=sum(latencies) / len(latencies) if latencies else 0,
+            min_latency_ms=min(latencies) if latencies else 0,
+            max_latency_ms=max(latencies) if latencies else 0,
+            total_cost_usd=total_cost,
+            total_tokens=total_tokens,
+        )
 
     def filter(
         self,
@@ -49,7 +47,7 @@ class Report:
         if passed is not None:
             filtered = [r for r in filtered if r.passed == passed]
 
-        return Report(results=filtered, metadata=self.metadata)
+        return Report(results=filtered)
 
     def group_by(self, key: Callable[[TestResult], str]) -> dict[str, "Report"]:
         groups: dict[str, list[TestResult]] = {}
@@ -60,16 +58,4 @@ class Report:
                 groups[group_key] = []
             groups[group_key].append(result)
 
-        return {k: Report(results=v, metadata=self.metadata) for k, v in groups.items()}
-
-    def compare_models(self) -> dict[str, dict[str, Any]]:
-        by_model = self.group_by(lambda r: r.test_case.model)
-        return {model: report.get_summary() for model, report in by_model.items()}
-
-    @staticmethod
-    def _percentile(values: list[float], p: float) -> float:
-        if not values:
-            return 0
-        sorted_values = sorted(values)
-        index = int(len(sorted_values) * p)
-        return sorted_values[min(index, len(sorted_values) - 1)]
+        return {k: Report(results=v) for k, v in groups.items()}

{promptum-0.0.1/src/promptum/core → promptum-0.0.3/src/promptum/benchmark}/result.py

@@ -2,8 +2,8 @@ from dataclasses import dataclass, field
 from datetime import datetime
 from typing import Any
 
-from promptum.core.metrics import Metrics
-from promptum.core.test_case import TestCase
+from promptum.benchmark.test_case import TestCase
+from promptum.providers.metrics import Metrics
 
 
 @dataclass(frozen=True, slots=True)

{promptum-0.0.1/src/promptum/execution → promptum-0.0.3/src/promptum/benchmark}/runner.py

@@ -3,8 +3,8 @@ from collections.abc import Callable, Sequence
 
 import httpx
 
-from promptum.core.result import TestResult
-from promptum.core.test_case import TestCase
+from promptum.benchmark.result import TestResult
+from promptum.benchmark.test_case import TestCase
 from promptum.providers.protocol import LLMProvider
 
 
@@ -37,7 +37,6 @@ class Runner:
 
         results = await asyncio.gather(
             *[run_with_semaphore(tc) for tc in test_cases],
-            return_exceptions=False,
         )
 
         return list(results)

promptum-0.0.3/src/promptum/benchmark/summary.py

@@ -0,0 +1,14 @@
+from dataclasses import dataclass
+
+
+@dataclass(frozen=True, slots=True)
+class Summary:
+    total: int
+    passed: int
+    failed: int
+    pass_rate: float
+    avg_latency_ms: float
+    min_latency_ms: float
+    max_latency_ms: float
+    total_cost_usd: float
+    total_tokens: int

{promptum-0.0.1/src/promptum/core → promptum-0.0.3/src/promptum/benchmark}/test_case.py

@@ -5,7 +5,7 @@ from typing import TYPE_CHECKING, Any
 if TYPE_CHECKING:
     from promptum.validation.protocol import Validator
 
-from promptum.core.retry import RetryConfig
+from promptum.providers.retry import RetryConfig
 
 
 @dataclass(frozen=True, slots=True)