repo stringclasses 454
values | file_path stringlengths 5 201 | extension stringclasses 1
value | content stringlengths 8 509k | num_lines int64 3 16.9k | size_bytes int64 8 511k |
|---|---|---|---|---|---|
cs249r_book | tinytorch/tests/16_compression/test_compression_integration.py | .py | #!/usr/bin/env python3
"""
Integration tests for Module 16: Compression
Tests pruning, knowledge distillation, and model compression
"""
import sys
from pathlib import Path
sys.path.insert(0, str(Path(__file__).parent.parent.parent))
def test_compression_integration():
"""Test compression system integration."""
... | 24 | 697 |
cs249r_book | tinytorch/tests/09_convolutions/test_convolutions_gradient_flow.py | .py | """
Test gradient flow through spatial operations (Conv2d, MaxPool2d).
These tests ensure that:
1. Conv2dBackward is properly attached to Conv2d outputs
2. MaxPool2dBackward is properly attached to MaxPool2d outputs
3. Gradients flow correctly to all parameters (weight, bias)
4. Integration with autograd system works ... | 301 | 10,077 |
cs249r_book | tinytorch/tests/09_convolutions/test_convolutions_core.py | .py | """
Module 09: Convolutions - Core Functionality Tests
===================================================
These tests verify convolutional layers work correctly for computer vision.
WHY CONVOLUTIONS MATTER:
-----------------------
Convolutions are the foundation of computer vision:
- Image classification (ImageNet, ... | 363 | 12,495 |
cs249r_book | tinytorch/tests/09_convolutions/test_09_convolutions_progressive.py | .py | """
Module 09: Progressive Integration Tests
Tests that Module 09 (Convolutions/Spatial) works correctly AND that prior modules (01→08) still work.
DEPENDENCY CHAIN: 01_tensor → 02_activations → 03_layers → 04_losses → 05_dataloader → 06_autograd → 07_optimizers → 08_training → 09_convolutions
⚠️ IMPORTANT: This test... | 493 | 17,261 |
cs249r_book | tinytorch/tests/e2e/test_user_journey.py | .py | """
End-to-End User Journey Tests for TinyTorch
These tests simulate the complete student experience:
1. Fresh start (setup)
2. Module workflow (start → work → complete)
3. Progress tracking
4. Milestone unlocking
Run with:
pytest tests/e2e/test_user_journey.py -v
Categories:
-k quick # Fast CLI veri... | 432 | 15,971 |
cs249r_book | tinytorch/tests/e2e/conftest.py | .py | """
E2E Test Configuration
Registers pytest markers for categorizing tests by speed and purpose.
"""
import pytest
def pytest_configure(config):
"""Register custom markers for E2E tests."""
config.addinivalue_line("markers", "quick: Quick verification tests (~30s total)")
config.addinivalue_line("marker... | 18 | 694 |
cs249r_book | tinytorch/tests/05_dataloader/test_05_dataloader_progressive.py | .py | """
Module 05: Progressive Integration Tests
Tests that Module 05 (DataLoader) works correctly AND that Foundation tier (01→04) still works.
DEPENDENCY CHAIN: 01_tensor → 02_activations → 03_layers → 04_losses → 05_dataloader
🎯 WHAT THIS TESTS:
- Module 05: Dataset abstraction, batching, shuffling, data pipelines
- ... | 471 | 16,839 |
cs249r_book | tinytorch/tests/05_dataloader/test_dataloader_core.py | .py | """
Module 05: DataLoader - Core Functionality Tests
=================================================
WHY DATALOADER MATTERS:
----------------------
Real datasets don't fit in memory. DataLoader:
- Loads data in batches
- Shuffles for better training
- Enables parallel loading
WHAT STUDENTS LEARN:
------------------... | 121 | 4,176 |
cs249r_book | tinytorch/paper/scripts/benchmark_quick.py | .py | #!/usr/bin/env python3
"""
Quick benchmark for Table 3 - uses reasonable approximations for slow operations
"""
import time
import numpy as np
import torch
def time_op(func, warmup=2, runs=5):
"""Time an operation"""
for _ in range(warmup):
func()
times = []
for _ in range(runs):
start... | 135 | 4,705 |
cs249r_book | tinytorch/src/04_losses/04_losses.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 04: Losses ... | 1,729 | 63,399 |
cs249r_book | tinytorch/src/07_optimizers/07_optimizers.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.19.2
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 07: Optimiz... | 2,025 | 71,803 |
cs249r_book | tinytorch/src/10_tokenization/10_tokenization.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 10: Tokeniz... | 2,026 | 79,209 |
cs249r_book | tinytorch/src/15_quantization/15_quantization.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
#| default_exp perf.quantization
# %% [... | 2,399 | 103,571 |
cs249r_book | tinytorch/src/18_memoization/18_memoization.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 18: Memoiza... | 2,062 | 81,781 |
cs249r_book | tinytorch/src/02_activations/02_activations.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 02: Activat... | 1,184 | 38,208 |
cs249r_book | tinytorch/src/03_layers/03_layers.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 03: Layers ... | 1,468 | 57,672 |
cs249r_book | tinytorch/src/11_embeddings/11_embeddings.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 11: Embeddi... | 2,008 | 88,246 |
cs249r_book | tinytorch/src/14_profiling/14_profiling.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 14: Profili... | 2,506 | 101,229 |
cs249r_book | tinytorch/src/20_capstone/20_capstone.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 20: Capston... | 2,185 | 92,313 |
cs249r_book | tinytorch/src/13_transformers/13_transformers.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
#| default_exp core.transformers
#| expo... | 2,035 | 87,730 |
cs249r_book | tinytorch/src/08_training/08_training.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 08: Trainin... | 2,049 | 73,221 |
cs249r_book | tinytorch/src/12_attention/12_attention.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
#| default_exp core.attention
#| export
... | 1,551 | 61,780 |
cs249r_book | tinytorch/src/17_acceleration/17_acceleration.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
#| default_exp perf.acceleration
#| expo... | 1,526 | 61,596 |
cs249r_book | tinytorch/src/01_tensor/01_tensor.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 01: Tensor ... | 2,002 | 79,813 |
cs249r_book | tinytorch/src/19_benchmarking/19_benchmarking.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 19: Benchma... | 4,120 | 155,862 |
cs249r_book | tinytorch/src/06_autograd/06_autograd.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 06: Autogra... | 3,498 | 125,647 |
cs249r_book | tinytorch/src/16_compression/16_compression.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 16: Compres... | 1,971 | 79,358 |
cs249r_book | tinytorch/src/09_convolutions/09_convolutions.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
# %% [markdown]
"""
# Module 09: Convolu... | 3,243 | 124,446 |
cs249r_book | tinytorch/src/05_dataloader/05_dataloader.py | .py | # ---
# jupyter:
# jupytext:
# text_representation:
# extension: .py
# format_name: percent
# format_version: '1.3'
# jupytext_version: 1.17.1
# kernelspec:
# display_name: Python 3 (ipykernel)
# language: python
# name: python3
# ---
#| default_exp core.dataloader
#| export... | 2,350 | 87,662 |
cs249r_book | slides/scripts/pdf2pptx.py | .py | #!/usr/bin/env python3
"""Convert PDF slide decks to PowerPoint (PPTX) using high-resolution images.
Each PDF page is rendered at 300 DPI via pdftoppm (poppler) and placed as a
full-bleed image on a 16:9 PowerPoint slide. The result is visually identical
to the PDF — suitable for presenting in PowerPoint/Keynote with ... | 125 | 4,280 |
cs249r_book | mlsysim/generate_appendix.py | .py | # generate_appendix.py
"""
mlsysim Appendix Generator
==========================
Generates Quarto-compatible Markdown tables for the textbook's backmatter.
Extracts live data from the mlsysim Hardware and Model registries.
"""
from mlsysim.core.units import Q_
from mlsysim.hardware.registry import Hardware
from mlsysi... | 53 | 1,956 |
cs249r_book | mlsysim/mlsysim/solvers.py | .py | """
mlsysim.solvers — Convenience re-export of all solver classes.
This is the stable public import path for solvers:
from mlsysim.solvers import ServingModel, TailLatencyModel, ...
The export list is derived mechanically from
``mlsysim.engine.solvers.__all__`` (the canonical list), so every name here is
``is``-... | 18 | 601 |
cs249r_book | mlsysim/mlsysim/__main__.py | .py | """Entry point for `python -m mlsysim`."""
try:
from mlsysim.cli.main import app
except ImportError:
import sys
print(
"Unable to import the mlsysim CLI.\n"
"Install or repair the package with: pip install mlsysim",
file=sys.stderr,
)
sys.exit(1)
if __name__ == "__main__":
... | 16 | 330 |
cs249r_book | mlsysim/mlsysim/__init__.py | .py | # mlsysim/__init__.py
"""
mlsysim: Machine Learning Systems Infrastructure and Modeling Platform
"""
__version__ = "0.1.2"
from . import core
from . import engine
from . import hardware
from . import models
from . import platforms
from . import infrastructure
from . import systems
from . import sim
from . import phys... | 75 | 2,631 |
cs249r_book | mlsysim/mlsysim/fmt.py | .py | """
fmt.py
Formatting + presentation helpers for Markdown/Quarto output.
Keep science in mlsysim/physics/; keep display here.
"""
from .core.units import ureg
class MarkdownStr(str):
"""A string that ALSO renders as raw Markdown when consumed by Quarto/Jupyter.
Quarto's inline ``{python} x`` substitution es... | 3,131 | 107,302 |
cs249r_book | mlsysim/mlsysim/show.py | .py | """
show.py — Tutorial display helpers for MLSys·im.
Replaces verbose print(f"...") patterns with clean, aligned output.
Two primitives: info() for key-value blocks, table() for tabular data.
Usage in tutorials:
from mlsysim.show import info, table, banner
info("Phase Analysis",
TTFT=result.ttft.to(... | 154 | 4,502 |
cs249r_book | mlsysim/mlsysim/systems/types.py | .py | from pydantic import BaseModel, ConfigDict, Field, field_validator
from typing import Any, Optional
from ..hardware.types import HardwareNode
from ..infrastructure.types import Datacenter, GridProfile
from ..core.units import ureg
from ..core.types import (
Quantity,
Metadata,
require_dimensionality,
re... | 242 | 8,265 |
cs249r_book | mlsysim/mlsysim/systems/reliability.py | .py | """Component MTTF and recovery assumptions (fleet reliability appendix)."""
from pydantic import BaseModel, ConfigDict, field_validator
from ..core.provenance import Sourced, sourced, fleet_mttf_hours
from ..core.registry import Registry
from ..core import provenance_catalog as pc
class ReliabilityComponent(BaseMod... | 148 | 6,020 |
cs249r_book | mlsysim/mlsysim/systems/registry.py | .py | from .types import (
CheckpointStoragePath,
Fleet,
NetworkFabric,
Node,
NodeStorageConfig,
PodEnvelope,
RackProfile,
StorageSubsystem,
)
from .reliability import Reliability
from .orchestration import Orchestration as OrchestrationProfile
from ..core.units import ureg, Q_, Gbps, GB, TB, ... | 432 | 17,585 |
cs249r_book | mlsysim/mlsysim/systems/orchestration.py | .py | """Fleet orchestration scenario parameters (queueing, utilization)."""
from pydantic import BaseModel, ConfigDict, Field
from ..core.types import Metadata
class Orchestration(BaseModel):
"""Shared cluster scheduling assumptions for scenario calculations."""
model_config = ConfigDict(arbitrary_types_allowed... | 16 | 525 |
cs249r_book | mlsysim/mlsysim/physics/transformer.py | .py | """Transformer FLOP accounting identities."""
from __future__ import annotations
from mlsysim.core.units import ureg
from mlsysim.literature.registry import Literature
from ._units import _ensure_unit
def calc_transformer_training_flops(n_params, n_tokens):
"""Training FLOPs for a Transformer (6PD rule, Kaplan... | 30 | 1,253 |
cs249r_book | mlsysim/mlsysim/physics/reliability.py | .py | """Reliability, checkpointing, and availability models."""
from __future__ import annotations
import math
from mlsysim.core.units import ureg
from mlsysim.core._validation import (
validate_positive,
validate_nonnegative,
validate_range,
validate_at_least,
)
from ._units import _ensure_unit
def ca... | 171 | 6,069 |
cs249r_book | mlsysim/mlsysim/physics/_units.py | .py | """Shared unit helpers for physics formulas."""
from __future__ import annotations
import pint
from mlsysim.core.units import ureg
def _ensure_unit(val, expected_unit, param_name="Value"):
"""
Attach a unit if val is a raw number; verify dimensional correctness AND
convert to ``expected_unit`` if it is... | 36 | 1,272 |
cs249r_book | mlsysim/mlsysim/physics/__init__.py | .py | """
Canonical physics and accounting formulas for ML systems.
Domain modules:
networking, performance, economics, memory, communication,
reliability, transformer, serving, statistics
"""
from ._units import _ensure_unit
from .constants import SPEED_OF_LIGHT_FIBER_KM_S
from .networking import calc_network_latency_... | 126 | 3,632 |
cs249r_book | mlsysim/mlsysim/physics/constants.py | .py | """Universal physical constants (the physics layer's ground truth).
These are genuine constants of nature / physical media — not hardware specs,
model specs, or tunable knobs — so they live with the laws in ``physics/``
rather than in ``core/constants.py`` (which is now units-only). See
the project MLSysIM rules -> Ca... | 14 | 670 |
cs249r_book | mlsysim/mlsysim/physics/quantities.py | .py | """Quantity-first formula helpers for LEGO cells — return Pint quantities, never strings."""
from __future__ import annotations
import pint
from mlsysim.core.units import (
Bparam,
byte,
count,
gram,
hour,
joule,
kilogram,
kWh,
param,
second,
ureg,
watt,
)
__all__ = [... | 112 | 3,843 |
cs249r_book | mlsysim/mlsysim/physics/networking.py | .py | """Network latency and distance physics."""
from __future__ import annotations
from mlsysim.core.units import ureg
from .constants import SPEED_OF_LIGHT_FIBER_KM_S
from ._units import _ensure_unit
def calc_network_latency_ms(distance_km):
"""Physical round-trip latency floor for a fiber link.
Models only ... | 34 | 1,105 |
cs249r_book | mlsysim/mlsysim/physics/communication.py | .py | """Collective communication time models (α–β).
All collectives here are pure communication models: the local reduction
compute term (γ in Thakur et al. 2005) is deliberately omitted, matching the
book's α–β pedagogical treatment.
"""
from __future__ import annotations
import math
from mlsysim.core.units import ureg... | 421 | 15,588 |
cs249r_book | mlsysim/mlsysim/physics/economics.py | .py | """Fleet economics and cloud cost models."""
from __future__ import annotations
from mlsysim.core.units import ureg, DAYS_PER_YEAR
from mlsysim.core._validation import validate_nonnegative, validate_positive
from ._units import _ensure_unit
def calc_monthly_egress_cost(bytes_per_sec, cost_per_gb):
"""
Calc... | 90 | 3,373 |
cs249r_book | mlsysim/mlsysim/physics/statistics.py | .py | """Statistical and workflow propagation helpers."""
from __future__ import annotations
import math
def calc_population_stability_index(expected, actual, epsilon=1e-12):
"""Population Stability Index (PSI) between two aligned distributions.
Measures distribution drift between a reference ("expected") and an... | 91 | 3,471 |
cs249r_book | mlsysim/mlsysim/physics/memory.py | .py | """Model and activation memory accounting."""
from __future__ import annotations
import math
import pint
from mlsysim.core.units import ureg, MB
from mlsysim.core._validation import validate_at_least, validate_nonnegative
from ._units import _ensure_unit
def model_memory(params, bytes_per_param, unit=MB):
"""... | 276 | 9,606 |
cs249r_book | mlsysim/mlsysim/physics/performance.py | .py | """Training time, scaling, roofline, and pipeline performance."""
from __future__ import annotations
from mlsysim.core.units import ureg
from mlsysim.core._validation import validate_positive, validate_at_least, validate_range
def dTime(total_ops, num_devices, peak_flops_per_device, efficiency_eta):
"""
Cor... | 232 | 8,481 |
cs249r_book | mlsysim/mlsysim/physics/serving.py | .py | """Inference serving and queueing models."""
from __future__ import annotations
import math
from mlsysim.core.units import ureg
from ._units import _ensure_unit
def calc_queue_latency_mmc(arrival_rate_hz, service_rate_hz, num_servers):
"""
M/M/c queueing model for inference tail latency (Erlang C).
C... | 88 | 4,125 |
cs249r_book | mlsysim/mlsysim/tools/audit_provenance.py | .py | #!/usr/bin/env python3
"""Report missing or weak provenance on registry entries."""
from __future__ import annotations
import argparse
import sys
from datetime import date
from typing import Any, Iterable
from mlsysim.core.provenance import Provenance, ProvenanceKind, Sourced
from mlsysim.datasets.registry import Da... | 361 | 12,439 |
cs249r_book | mlsysim/mlsysim/engine/pipeline.py | .py | """Pipeline composer for chaining mlsysim analytical models and solvers.
Layer C of the composition architecture: a transparent Pipeline that
chains resolvers (models and solvers), validates compatibility, and
shows students the full Demand → Supply → Consequence data flow.
The Pipeline is NOT a black box — it is an... | 175 | 6,779 |
cs249r_book | mlsysim/mlsysim/engine/evaluation.py | .py | from pydantic import BaseModel, ConfigDict
from typing import Optional, Dict, Any
class EvaluationLevel(BaseModel):
"""A single tier in the Hierarchy of Constraints."""
level_name: str
status: str = "PASS" # PASS, FAIL, WARNING
summary: str
metrics: Dict[str, Any] = {}
class SystemEvaluation(BaseM... | 242 | 9,952 |
cs249r_book | mlsysim/mlsysim/engine/empirical.py | .py | """Domain-reviewed empirical anchors for MLSysIM sanity checks.
These anchors are not solver defaults and do not define model or hardware
facts. They bind canonical ``Models.*`` and ``Hardware.*`` entries to sourced
``Literature.Benchmarks`` values so tests can catch formula or registry drift
without duplicating the u... | 191 | 7,068 |
cs249r_book | mlsysim/mlsysim/engine/walls.py | .py | """The 22 ML Systems Walls — canonical taxonomy.
This module is the single source of truth for the wall classification
used throughout MLSysIM analyses, papers, and notebooks. Every wall
represents a physical or logical constraint that bounds system
performance; each is resolved by a dedicated solver.
The walls are o... | 396 | 13,570 |
cs249r_book | mlsysim/mlsysim/engine/calibration.py | .py | """Parameters for analytical solvers and the roofline engine.
These values tune ``mlsysim.engine.solvers`` models and ``mlsysim.engine.engine.Engine``
when callers omit explicit arguments. Use ``Literature.*``, ``Systems.*``, and
``Infrastructure.*`` for sourced domain reference values.
"""
from ..core.provenance imp... | 73 | 3,404 |
cs249r_book | mlsysim/mlsysim/engine/scenarios.py | .py | from pydantic import BaseModel, ConfigDict
from typing import Optional, Union
from ..core.units import Q_
from ..core.types import Quantity
from ..models.types import Workload
from ..hardware.types import HardwareNode
from ..systems.types import Fleet
from ..core.exceptions import OOMError, SLAViolation
from .evaluatio... | 273 | 10,619 |
cs249r_book | mlsysim/mlsysim/engine/engine.py | .py | from pydantic import BaseModel, ConfigDict, Field
from typing import Optional, List
from ..core.units import ureg, Q_, resolve_precision
from . import calibration as cal
from ..physics import calc_bottleneck
from ..core.exceptions import OOMError
from ..core._validation import validate_range, validate_at_least
from ..m... | 401 | 19,107 |
cs249r_book | mlsysim/mlsysim/engine/resolver_factory.py | .py | import logging
from typing import Type, Dict
from .solvers import BaseResolver
logger = logging.getLogger(__name__)
class ResolverFactory:
"""
Factory for creating and retrieving Solvers/Models.
This acts as the entry point for the Pluggable Solvers interface.
It automatically discovers all built... | 81 | 3,024 |
cs249r_book | mlsysim/mlsysim/engine/explainers.py | .py | import logging
from typing import Any
logger = logging.getLogger(__name__)
class DifferentialExplainer:
"""
Automates the 'Why did this happen?' analysis by comparing two solver results.
It identifies the binding constraints and mathematically explains the performance delta.
"""
@staticmethod... | 68 | 3,747 |
cs249r_book | mlsysim/mlsysim/engine/config.py | .py | from pydantic import BaseModel, ConfigDict, Field, model_validator
from typing import Dict, Any
from ..models.registry import Models
from ..hardware.registry import Hardware
class SimulationConfig(BaseModel):
"""
Standard schema for an ML systems analytical modeling run.
Can be loaded from YAML, JSON, or P... | 56 | 2,022 |
cs249r_book | mlsysim/mlsysim/engine/dse.py | .py | from typing import Dict, Any, List, Optional, Callable
from pydantic import BaseModel, Field
import logging
import itertools
logger = logging.getLogger(__name__)
class SearchSpace(BaseModel):
"""
Defines the dimensions and discrete bounds of the design space.
Example: {"tp": [1, 2, 4, 8], "pp": [1, 2, 4],... | 165 | 6,706 |
cs249r_book | mlsysim/mlsysim/engine/results.py | .py | """Typed result models for all mlsysim models and solvers.
Layer A of the composition architecture: every resolver returns a typed
Pydantic model instead of Dict[str, Any]. This gives students
autocomplete, documentation, and type safety when composing analytical
models and analysis solvers.
"""
from __future__ imp... | 413 | 12,758 |
cs249r_book | mlsysim/mlsysim/engine/solvers/reliability.py | .py | """Reliability and checkpoint-interval solvers.
Domain implementations behind ``mlsysim.solvers`` (the public import
path, derived from ``engine.solvers.__init__``); kept per-domain so the logic stays reviewable.
"""
from __future__ import annotations
from ..results import (
ReliabilityResult,
)
from ...physics... | 101 | 4,076 |
cs249r_book | mlsysim/mlsysim/engine/solvers/utils.py | .py | from __future__ import annotations
from ...systems.types import NetworkFabric, Node
def _intra_node_latency(node: Node):
"""Resolve the per-hop latency for intra-node (NVLink) communication.
Implements the instance -> tech-class fallback: prefer the latency on the
accelerator's own NVLink spec (instance... | 57 | 1,827 |
cs249r_book | mlsysim/mlsysim/engine/solvers/distributed.py | .py | """Distributed training, routing, topology, and parallelism search solvers.
Domain implementations behind ``mlsysim.solvers`` (the public import
path, derived from ``engine.solvers.__init__``); kept per-domain so the logic stays reviewable.
"""
from __future__ import annotations
import math
from typing import Option... | 711 | 32,639 |
cs249r_book | mlsysim/mlsysim/engine/solvers/__init__.py | .py | """Domain-oriented MLSysIM solver implementations."""
from .base import BaseOptimizer, BaseResolver, BaseSolver, ForwardModel
from .compression import CompressionModel
from .data import DataModel, TransformationModel
from .distributed import DistributedModel, MoERoutingModel, ParallelismOptimizer, TopologyModel
from .... | 56 | 1,690 |
cs249r_book | mlsysim/mlsysim/engine/solvers/training.py | .py | """Training memory, checkpointing, and scaling-law solvers.
Domain implementations behind ``mlsysim.solvers`` (the public import
path, derived from ``engine.solvers.__init__``); kept per-domain so the logic stays reviewable.
"""
from __future__ import annotations
import math
from typing import Optional
from ..resul... | 357 | 16,165 |
cs249r_book | mlsysim/mlsysim/engine/solvers/data.py | .py | """Data-ingestion and preprocessing pipeline solvers.
Domain implementations behind ``mlsysim.solvers`` (the public import
path, derived from ``engine.solvers.__init__``); kept per-domain so the logic stays reviewable.
"""
from __future__ import annotations
from ..results import (
DataResult,
Transformation... | 150 | 6,072 |
cs249r_book | mlsysim/mlsysim/engine/solvers/orchestration.py | .py | """Cluster orchestration and queueing solvers.
Domain implementations behind ``mlsysim.solvers`` (the public import
path, derived from ``engine.solvers.__init__``); kept per-domain so the logic stays reviewable.
"""
from __future__ import annotations
from ..results import (
OrchestrationResult,
)
from ...core.u... | 83 | 3,014 |
cs249r_book | mlsysim/mlsysim/engine/solvers/compression.py | .py | """Model compression trade-off solvers.
Domain implementations behind ``mlsysim.solvers`` (the public import
path, derived from ``engine.solvers.__init__``); kept per-domain so the logic stays reviewable.
"""
from __future__ import annotations
import math
from typing import Any, Dict, List, Optional
from ..results ... | 387 | 16,915 |
cs249r_book | mlsysim/mlsysim/engine/solvers/economics.py | .py | """Sustainability, economics, responsible-engineering, and placement solvers.
Domain implementations behind ``mlsysim.solvers`` (the public import
path, derived from ``engine.solvers.__init__``); kept per-domain so the logic stays reviewable.
"""
from __future__ import annotations
from typing import Any, List, Optio... | 425 | 18,721 |
cs249r_book | mlsysim/mlsysim/engine/solvers/base.py | .py | from __future__ import annotations
from abc import ABC, abstractmethod
from typing import Any, Dict, Optional, Type
from ..results import SolverResult
class BaseResolver(ABC):
"""Base class for all mlsysim analytical components (Models, Solvers, Optimizers).
Each resolver declares its input requirements and... | 88 | 2,953 |
cs249r_book | mlsysim/mlsysim/engine/solvers/performance.py | .py | """Single-node, network-roofline, efficiency, and inverse-design solvers.
Domain implementations behind ``mlsysim.solvers`` (the public import
path, derived from ``engine.solvers.__init__``); kept per-domain so the logic stays reviewable.
"""
from __future__ import annotations
from ..engine import Engine, Performan... | 457 | 20,418 |
cs249r_book | mlsysim/mlsysim/engine/solvers/serving.py | .py | """LLM serving, batching, tail-latency, and inference-scaling solvers.
Domain implementations behind ``mlsysim.solvers`` (the public import
path, derived from ``engine.solvers.__init__``); kept per-domain so the logic stays reviewable.
"""
from __future__ import annotations
import math
from typing import Optional
f... | 1,011 | 49,940 |
cs249r_book | mlsysim/mlsysim/datasets/types.py | .py | from pydantic import BaseModel, ConfigDict, Field, field_validator
from typing import Optional
from ..core.units import ureg
from ..core.types import Metadata, Quantity, require_dimensionality, require_unit_family
class DatasetProfile(BaseModel):
model_config = ConfigDict(arbitrary_types_allowed=True, extra="for... | 45 | 1,827 |
cs249r_book | mlsysim/mlsysim/datasets/__init__.py | .py | """Dataset zoo — canonical data corpus profiles."""
from .registry import Datasets
from .types import DatasetProfile
__all__ = ["Datasets", "DatasetProfile"]
| 7 | 162 |
cs249r_book | mlsysim/mlsysim/datasets/registry.py | .py | """Dataset registry — dataset profiles.
Leaf reference data (example counts, image dimensions, class counts) lives as
YAML under ``datasets/data/<Dataset>.yaml`` and is loaded + validated against
the ``DatasetProfile`` schema at import (see ``core/loader.py`` and
the project MLSysIM rules → *Storage format*).
"""
from... | 19 | 577 |
cs249r_book | mlsysim/mlsysim/platforms/types.py | .py | from pydantic import BaseModel, ConfigDict, Field, field_validator
from ..core.units import ureg
from ..core.types import Metadata, Quantity, require_dimensionality, require_unit_family
class PlatformEnvelope(BaseModel):
"""Abstract deployment envelope (RAM, storage, latency budget)."""
model_config = Confi... | 58 | 1,948 |
cs249r_book | mlsysim/mlsysim/platforms/__init__.py | .py | """Platform deployment envelopes."""
from .registry import Platforms
from .types import PlatformEnvelope
__all__ = ["Platforms", "PlatformEnvelope"]
| 7 | 151 |
cs249r_book | mlsysim/mlsysim/platforms/registry.py | .py | """Deployment paradigm envelopes (Cloud, Edge, Mobile, TinyML)."""
from ..core.units import (
GB,
GiB,
KiB,
MB,
PFLOP,
TB,
TFLOPs,
TOPS,
milliwatt,
second,
ureg,
)
from ..core.registry import Registry
from ..core.types import Metadata
from ..core import provenance_catalog as... | 86 | 2,561 |
cs249r_book | mlsysim/mlsysim/viz/plots.py | .py | # viz/plots.py
# Simulator-aware plotters for generated MLSysIM figures.
# Book publication style is owned by book.tools.figures.style; this module keeps
# a local fallback so standalone mlsysim installs do not depend on the book tree.
import os
try:
import numpy as np
except ImportError:
np = None
try:
... | 492 | 15,117 |
cs249r_book | mlsysim/mlsysim/core/types.py | .py | from typing import Any, Annotated, Optional
from pydantic import AfterValidator, PlainSerializer, BaseModel, ConfigDict
from .units import Q_
from .provenance import Provenance
def validate_quantity(v: Any) -> Q_:
if isinstance(v, Q_):
return v
if isinstance(v, (int, float, str)):
try:
... | 119 | 4,269 |
cs249r_book | mlsysim/mlsysim/core/exceptions.py | .py | # Exceptions for the MLSys Simulator
class MLSysError(Exception):
"""Base exception for all mlsysim simulation errors."""
pass
class OOMError(MLSysError):
"""Raised when a workload's memory footprint exceeds the hardware capacity."""
def __init__(self, message, required_bytes=None, available_bytes=Non... | 21 | 723 |
cs249r_book | mlsysim/mlsysim/core/_validation.py | .py | """Input validation helpers for mlsysim formulas and solvers.
These guards catch common student mistakes (zero bandwidth, negative efficiency,
n_gpus=0) before they produce confusing inf/nan results or crash with unhelpful
error messages.
"""
def validate_positive(val, name: str):
"""Ensure a numeric value is st... | 35 | 1,163 |
cs249r_book | mlsysim/mlsysim/core/provenance.py | .py | """Provenance types for registry entries and public sourced scalars."""
from __future__ import annotations
from datetime import date
from enum import Enum
from typing import Optional, Union
from pydantic import BaseModel, ConfigDict, Field, model_validator
Scalar = Union[int, float]
class ProvenanceKind(str, Enum... | 199 | 7,232 |
cs249r_book | mlsysim/mlsysim/core/loader.py | .py | """YAML data-layer loader.
Leaf reference data (hardware chips, models, datasets, …) lives as YAML and is
loaded + validated against a pydantic schema at import, then assembled into a
``Registry`` subclass whose consumer API is identical to the former
hand-written Python registry (``Hardware.Cloud.H100.memory.bandwidt... | 214 | 9,000 |
cs249r_book | mlsysim/mlsysim/core/units.py | .py | # units.py
# Measurement infrastructure for the ML Systems simulator.
# This module owns the pint unit registry and defines all unit aliases.
# It contains ONLY measurement plumbing — no domain knowledge or tuneable defaults.
from pathlib import Path
import pint
__all__ = [
# Registry and Quantity constructor
... | 275 | 8,303 |
cs249r_book | mlsysim/mlsysim/core/provenance_catalog.py | .py | """Shared provenance records (stable ids, single definition)."""
from __future__ import annotations
from .provenance import Provenance, ProvenanceKind
def _ds(
id: str,
ref: str,
url: str,
*,
verified: str = "2026-03-06",
notes: str | None = None,
) -> Provenance:
"""Creates a Provenance... | 961 | 35,739 |
cs249r_book | mlsysim/mlsysim/core/registry/plugin_manager.py | .py | import logging
from typing import Dict, Any, TypeVar
import importlib.metadata
logger = logging.getLogger(__name__)
T = TypeVar('T')
class Registry:
"""
A generic plugin registry that dynamically discovers and loads classes or constants
from Python entry_points, allowing third parties to inject custom ha... | 61 | 2,180 |
cs249r_book | mlsysim/mlsysim/core/registry/__init__.py | .py | from typing import ClassVar, List, Any, Optional
from .plugin_manager import hardware_registry, model_registry, constants_registry
from .plugin_manager import Registry as PluginRegistry
class Registry:
"""
Base class for registries that provides a coherent way to list and sort items.
Used by Hardware, Mo... | 75 | 2,462 |
cs249r_book | mlsysim/mlsysim/core/optimization/__init__.py | .py | from .protocol import OptimizerProtocol, OptimizationResult
__all__ = [
"OptimizerProtocol",
"OptimizationResult",
"OptimizationRegistry",
]
def __getattr__(name):
if name == "OptimizationRegistry":
from .registry import OptimizationRegistry
return OptimizationRegistry
raise Attri... | 15 | 380 |
cs249r_book | mlsysim/mlsysim/core/optimization/scipy_backend.py | .py | import time
from typing import Any, Callable, Optional, Tuple
import scipy.optimize
from .protocol import OptimizerProtocol, OptimizationResult
class ScipyBackend(OptimizerProtocol):
"""
A continuous optimization backend using SciPy.
Best suited for finding optimal continuous system variables
(like co... | 74 | 2,974 |
cs249r_book | mlsysim/mlsysim/core/optimization/registry.py | .py | from .protocol import OptimizerProtocol
def _load_scipy_backend():
"""Lazy-loads the SciPy optimization backend."""
try:
from .scipy_backend import ScipyBackend
return ScipyBackend
except ImportError:
raise ImportError(
"SciPy is required for continuous optimization. "
... | 61 | 2,101 |
cs249r_book | mlsysim/mlsysim/core/optimization/protocol.py | .py | from typing import Protocol, Any, Dict
from pydantic import BaseModel
class OptimizationResult(BaseModel):
"""Standardized output from any solver backend (SciPy, OR-Tools, etc.)"""
feasible: bool
optimal_value: float
best_configuration: Dict[str, Any]
metrics: Dict[str, Any]
solver_name: str
... | 23 | 742 |
cs249r_book | mlsysim/mlsysim/core/optimization/exhaustive_backend.py | .py | import time
from typing import Any, Callable, List, Tuple
import numpy as np
from .protocol import OptimizerProtocol, OptimizationResult
class ExhaustiveBackend(OptimizerProtocol):
"""
A brute-force grid search backend using only NumPy.
Evaluates the objective at every point on a uniform grid and returns... | 77 | 3,019 |
cs249r_book | mlsysim/mlsysim/core/optimization/ortools_backend.py | .py | import time
from typing import Any, Callable, Dict
from ortools.sat.python import cp_model
from .protocol import OptimizerProtocol, OptimizationResult
class ORToolsDiscreteBackend(OptimizerProtocol):
"""
A discrete optimization backend using Google OR-Tools CP-SAT.
Best suited for finding optimal integer c... | 78 | 3,250 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.