chore: release v0.71.0 — split heavy deps into [train] extra

Heavy training stack (torch, transformers, peft, trl, datasets,
bitsandbytes, accelerate) moves out of the core install into a new
[train] optional-dependency extra. `pip install soup-cli` is now a
light CLI + data-tools install with no PyTorch; `pip install
'soup-cli[train]'` adds the training stack.

- pyproject: new [train] + [all] extras; [dev] self-references [train]
  so CI (`pip install -e ".[dev]"`) still gets torch. Pins unchanged.
- errors.py: missing torch/transformers/peft/trl/datasets/bitsandbytes/
  accelerate now surface a single 'install soup-cli[train]' fix.
- Dockerfile: install soup-cli[train,serve,data,eval] so the GPU image
  can still fine-tune.
- README + docs/models.md: split install into light core vs [train].
- CHANGELOG: cut [0.71.0]; bump version 0.70.0 -> 0.71.0.

CHANGELOG.md

@@ -12,11 +12,28 @@ reproducing 70+ versions of notes.
 
 ## [Unreleased]
 
+## [0.71.0] - 2026-06-01
+
 ### Changed
+- **Breaking — install split.** The heavy training stack (`torch`,
+  `transformers`, `peft`, `trl`, `datasets`, `bitsandbytes`, `accelerate`) moved
+  out of the core install into a new `[train]` extra. `pip install soup-cli` is
+  now a light CLI + data-tools install with **no PyTorch**; run
+  `pip install 'soup-cli[train]'` (or `[all]`) to fine-tune. Existing users who
+  train must reinstall with `[train]`. Version pins are unchanged.
+- Trimmed `README.md` to a ~238-line front door; the full feature reference now
+  lives under `docs/` (one topic page per area, indexed from the README).
+- Raised the pytest coverage gate from 50% to 77% (`--cov-fail-under=77`).
 - Migrated to a `src/` layout (`src/soup_cli/`) for cleaner packaging and to
   stop tests accidentally importing the in-tree package.
 
 ### Added
+- `[train]` and `[all]` optional-dependency extras (`[all]` pulls
+  `train`, `serve`, `ui`, `data`). `[dev]` self-references `[train]` so CI and
+  contributors still get the full stack from `pip install -e ".[dev]"`.
+- Friendly error mapping: a missing heavy dependency (`torch`, `transformers`,
+  `peft`, `trl`, `datasets`, `bitsandbytes`, `accelerate`) now surfaces
+  "Training needs the [train] extra. Run: pip install 'soup-cli[train]'".
 - `py.typed` marker (PEP 561) so downstream type checkers pick up Soup's inline
   type hints.
 - `.pre-commit-config.yaml` with ruff (lint + format) and standard file-hygiene
@@ -29,4 +46,5 @@ reproducing 70+ versions of notes.
   `SECURITY.md` (~220 KB). `SECURITY.md` is now a concise security policy; the
   detailed hardening notes remain in git history and the GitHub Releases notes.
 
-[Unreleased]: https://github.com/MakazhanAlpamys/Soup/compare/v0.70.0...HEAD
+[Unreleased]: https://github.com/MakazhanAlpamys/Soup/compare/v0.71.0...HEAD
+[0.71.0]: https://github.com/MakazhanAlpamys/Soup/compare/v0.70.0...v0.71.0

Dockerfile

@@ -23,8 +23,10 @@ RUN update-alternatives --install /usr/bin/python python /usr/bin/python3 1
 # Set working directory
 WORKDIR /workspace
 
-# Install Soup from PyPI (always the latest published release, not local source)
-RUN pip install --no-cache-dir "soup-cli[serve,data,eval]"
+# Install Soup from PyPI (always the latest published release, not local source).
+# v0.71.0 split the training stack into the [train] extra — include it so the
+# GPU image can still fine-tune.
+RUN pip install --no-cache-dir "soup-cli[train,serve,data,eval]"
 
 # Default entrypoint and command
 ENTRYPOINT ["soup"]

README.md

@@ -32,7 +32,7 @@
 Soup turns the pain of LLM fine-tuning into a simple workflow. One config, one command, done.
 
 ```bash
-pip install soup-cli
+pip install 'soup-cli[train]'   # add [train] to fine-tune; bare `soup-cli` is the light CLI
 soup init --template chat
 soup train
 ```
@@ -49,10 +49,10 @@ infrastructure instead of improving models. Soup fixes that.
 
 ## What's New
 
-**v0.70.0 — Loop Hardening.** Six surfaces that protect the training loop from the failure
-modes that cost a real GPU-hour: a reward-hacking detector (InfoRM / RM-ensemble), cross-tokenizer
-distillation (ULD), MiniLLM reverse-KL distillation, mid-epoch RL checkpoints, an iterative-DPO
-loop driver, and a RAGEN echo-trap detector. See [`docs/training.md`](docs/training.md).
+**v0.71.0 — Lighter install.** The heavy training stack (PyTorch, Transformers, PEFT, TRL,
+datasets, bitsandbytes, accelerate) moved into a `[train]` extra. `pip install soup-cli` is now a
+light CLI + data-tools install with no PyTorch; `pip install 'soup-cli[train]'` adds everything you
+need to fine-tune. **Breaking:** existing users who train must reinstall with `[train]`.
 
 Full history: [CHANGELOG.md](CHANGELOG.md) · [GitHub Releases](https://github.com/MakazhanAlpamys/Soup/releases).
 
@@ -61,10 +61,14 @@ Full history: [CHANGELOG.md](CHANGELOG.md) · [GitHub Releases](https://gi
 ### 1. Install
 
 ```bash
-pip install soup-cli                                   # from PyPI
+pip install soup-cli            # light: CLI + config + data tools (no PyTorch)
+pip install 'soup-cli[train]'   # add the training stack (torch, transformers, peft, trl, …)
 pip install git+https://github.com/MakazhanAlpamys/Soup.git   # latest dev
 ```
 
+`soup init`, `soup data …`, and the other data/inspection commands work on the light install.
+Fine-tuning (`soup train`) needs the `[train]` extra.
+
 ### 2. Create a config
 
 ```bash
@@ -203,8 +207,9 @@ docker compose up   # or build locally
 - 8 GB+ VRAM for 7B models with QLoRA
 
 All training tasks run on CPU for testing (quantization auto-disabled). Optional extras
-(`fast`, `vision`, `qat`, `serve`, `serve-fast`, `ui`, `eval`, `deepspeed`, `liger`, `mlx`,
-`onnx`, `tensorrt`, …) are listed in [`docs/models.md`](docs/models.md#optional-extras).
+(`train`, `all`, `fast`, `vision`, `qat`, `serve`, `serve-fast`, `ui`, `eval`, `deepspeed`,
+`liger`, `mlx`, `onnx`, `tensorrt`, …) are listed in
+[`docs/models.md`](docs/models.md#optional-extras).
 
 ## Troubleshooting
 

docs/models.md

@@ -50,10 +50,13 @@ Soup works with **any** of the **340,000+** text-generation models on [HuggingFa
 
 ## Optional Extras
 
-The core `pip install soup-cli` is enough to train. Install extras only when you need them:
+The core `pip install soup-cli` is a light install — the CLI, config system, and data tools, with
+no PyTorch. Add `[train]` to fine-tune, or install other extras only when you need them:
 
 | Extra | Install | What it adds |
 |---|---|---|
+| `train` | `pip install 'soup-cli[train]'` | Training stack: torch, transformers, peft, trl, datasets, bitsandbytes, accelerate |
+| `all` | `pip install 'soup-cli[all]'` | `train` + `serve` + `ui` + `data` in one shot |
 | `fast` | `pip install 'soup-cli[fast]'` | Unsloth backend (2-5x faster, lower VRAM) |
 | `vision` | `pip install 'soup-cli[vision]'` | Vision / multimodal fine-tuning (Pillow) |
 | `audio` | `pip install 'soup-cli[audio]'` | Audio / speech fine-tuning (librosa, soundfile) |

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "soup-cli"
-version = "0.70.0"
+version = "0.71.0"
 description = "Fine-tune LLMs in one command. No SSH, no config hell."
 readme = "README.md"
 license = "Apache-2.0"
@@ -21,11 +21,23 @@ classifiers = [
     "Programming Language :: Python :: 3",
     "Topic :: Scientific/Engineering :: Artificial Intelligence",
 ]
+# Core install (`pip install soup-cli`) is intentionally light: the CLI, config
+# system, and data tools — no PyTorch. v0.71.0 split the heavy training stack
+# (torch / transformers / peft / trl / datasets / bitsandbytes / accelerate)
+# into the `[train]` extra below.
 dependencies = [
     "typer>=0.9.0,<0.21.0",
     "rich>=13.0.0",
     "pydantic>=2.0.0",
     "pyyaml>=6.0",
+    "huggingface-hub>=0.16.0",
+    "plotext>=5.2.0",
+]
+
+[project.optional-dependencies]
+# v0.71.0 — heavy training stack. `pip install 'soup-cli[train]'` to fine-tune.
+# These were core dependencies through v0.70.0; pins are unchanged.
+train = [
     "torch>=2.0.0",
     "transformers>=4.36.0,<5.0.0",
     "peft>=0.7.0",
@@ -33,15 +45,16 @@ dependencies = [
     "datasets>=2.14.0",
     "bitsandbytes>=0.41.0",
     "accelerate>=0.25.0",
-    "huggingface-hub>=0.16.0",
-    "plotext>=5.2.0",
 ]
-
-[project.optional-dependencies]
+# v0.71.0 — convenience meta-extra pulling the main optional stacks.
+all = ["soup-cli[train,serve,ui,data]"]
 eval = ["lm-eval>=0.4.0"]
 data = ["datasketch>=1.6.0"]
 wandb = ["wandb>=0.15.0,<0.18.0"]
-dev = ["pytest>=7.0", "ruff>=0.1.0", "pytest-cov>=4.0", "httpx>=0.24.0", "mypy>=1.8.0", "pre-commit>=3.5.0"]
+# Self-references `[train]` so CI / contributors get the full training stack
+# (CI runs `pip install -e ".[dev]"`; without this every test would fail at
+# `import torch`).
+dev = ["soup-cli[train]", "pytest>=7.0", "ruff>=0.1.0", "pytest-cov>=4.0", "httpx>=0.24.0", "mypy>=1.8.0", "pre-commit>=3.5.0"]
 ui = ["fastapi>=0.104.0", "uvicorn>=0.24.0"]
 serve = ["fastapi>=0.104.0", "uvicorn>=0.24.0"]
 serve-fast = ["vllm>=0.4.0", "fastapi>=0.104.0", "uvicorn>=0.24.0"]

src/soup_cli/__init__.py

@@ -1,3 +1,3 @@
 """Soup CLI — Fine-tune LLMs in one command."""
 
-__version__ = "0.70.0"
+__version__ = "0.71.0"

src/soup_cli/utils/errors.py

@@ -7,6 +7,12 @@ from rich.panel import Panel
 
 console = Console(stderr=True)
 
+# v0.71.0 — the heavy training stack (torch / transformers / peft / trl /
+# datasets / bitsandbytes / accelerate) moved out of the core install into the
+# `[train]` extra. A missing one of these surfaces this single, actionable fix.
+# The `\\[` escapes the literal `[` for Rich markup (it renders as `[train]`).
+_TRAIN_FIX = "Training needs the \\[train] extra. Run: pip install 'soup-cli\\[train]'"
+
 # Map known error patterns to (short message, fix suggestion)
 ERROR_MAP = [
     # CUDA OOM
@@ -62,21 +68,41 @@ ERROR_MAP = [
         "httpx is not installed (needed for data generate).",
         "Run: pip install 'soup-cli\\[generate]'",
     ),
-    # Peft / transformers incompatibility
+    # Heavy training stack — all moved to the [train] extra in v0.71.0.
+    (
+        "No module named 'torch'",
+        "PyTorch is not installed (needed for training).",
+        _TRAIN_FIX,
+    ),
+    (
+        "No module named 'transformers'",
+        "Transformers is not installed (needed for training).",
+        _TRAIN_FIX,
+    ),
     (
         "No module named 'peft'",
-        "PEFT is not installed.",
-        "Run: pip install peft>=0.7.0",
+        "PEFT is not installed (needed for LoRA training).",
+        _TRAIN_FIX,
     ),
     (
         "No module named 'trl'",
-        "TRL is not installed.",
-        "Run: pip install trl>=0.7.0",
+        "TRL is not installed (needed for training).",
+        _TRAIN_FIX,
+    ),
+    (
+        "No module named 'datasets'",
+        "Datasets is not installed (needed for training).",
+        _TRAIN_FIX,
     ),
     (
         "No module named 'bitsandbytes'",
         "BitsAndBytes is not installed (needed for quantization).",
-        "Run: pip install bitsandbytes>=0.41.0",
+        _TRAIN_FIX,
+    ),
+    (
+        "No module named 'accelerate'",
+        "Accelerate is not installed (needed for training).",
+        _TRAIN_FIX,
     ),
     # CPU / quantization issues
     (

tests/test_errors.py

@@ -3,6 +3,7 @@
 from io import StringIO
 from unittest.mock import patch
 
+import pytest
 from rich.console import Console
 from typer.testing import CliRunner
 
@@ -72,6 +73,23 @@ def test_missing_deepspeed_error():
     assert "soup-cli[deepspeed]" in output
 
 
+@pytest.mark.parametrize(
+    "module",
+    ["torch", "transformers", "peft", "trl", "datasets", "bitsandbytes", "accelerate"],
+)
+def test_missing_train_dep_points_at_train_extra(module):
+    """v0.71.0 — any missing heavy training dep points at the [train] extra."""
+    buf = StringIO()
+    test_console = Console(file=buf, stderr=False)
+    with patch("soup_cli.utils.errors.console", test_console):
+        exc = ModuleNotFoundError(f"No module named '{module}'")
+        format_friendly_error(exc, verbose=False)
+    output = buf.getvalue()
+    # The `\[train]` escape renders to a literal `[train]` in the output.
+    assert "soup-cli[train]" in output
+    assert "[train] extra" in output
+
+
 def test_connection_error():
     """Connection error gets friendly message."""
     buf = StringIO()