unsloth/studio/backend/core/inference/diffusion.py

612 lines
25 KiB
Python

# SPDX-License-Identifier: AGPL-3.0-only
# Copyright 2026-present the Unsloth AI Inc. team. All rights reserved. See /studio/LICENSE.AGPL-3.0
"""Local diffusion (text-to-image) backend.
A torch-only singleton: it dequantises a single-file GGUF on-device via
``GGUFQuantizationConfig`` and pulls the rest of the pipeline (VAE, text
encoders, scheduler) from the matching base repo. torch/diffusers are imported
lazily so this stays importable in a no-torch runtime. ``begin_load`` runs on a
background thread; poll ``load_progress`` for the download bar. GPU-handoff
policy lives in the arbiter the routes call, not here.
"""
from __future__ import annotations
import inspect
import threading
import time
from dataclasses import dataclass
from pathlib import Path
from typing import Any, Optional
from loggers import get_logger
from utils.hardware import clear_gpu_cache
from .diffusion_families import (
DiffusionFamily,
detect_family,
resolve_base_repo,
resolve_local_gguf_child,
)
logger = get_logger(__name__)
@dataclass(frozen = True)
class _LoadState:
"""Everything about the currently-loaded pipeline, swapped as one unit."""
pipe: Any
family: Any
repo_id: str
base_repo: str
device: str
dtype: str
cpu_offload: bool
@dataclass
class _LoadingState:
"""An in-flight background load, polled for download progress."""
repo_id: str
base_repo: str
expected_bytes: int = 0
error: Optional[str] = None
@dataclass
class _GenState:
"""An in-flight generation, updated per denoising step for the progress bar."""
total_steps: int
step: int = 0
# Set when the first step finishes; the ETA rate is measured from there so the
# slower first step (warmup) doesn't skew it.
first_step_at: float = 0.0
# Computed once per step (in the callback) so it's stable between polls.
eta_seconds: Optional[float] = None
def _estimate_eta(total_steps: int, step: int, first_step_at: float, now: float) -> Optional[float]:
"""Seconds remaining, from the average step time measured after the first step.
None until at least one step has elapsed since the first."""
steps_since_first = step - 1
if not first_step_at or steps_since_first <= 0:
return None
per_step = (now - first_step_at) / steps_since_first
return max(0.0, (total_steps - step) * per_step)
class DiffusionBackend:
"""Holds at most one loaded diffusers pipeline. All mutations are serialised."""
def __init__(self) -> None:
# One lock serialises load / generate / unload — a generate must not run
# while the pipeline is being swapped out from under it. status() and
# load_progress() read the single state references without the lock, so
# polling never blocks a slow load.
self._lock = threading.Lock()
self._state: Optional[_LoadState] = None
self._loading: Optional[_LoadingState] = None
# Bumped on every begin_load and unload so a worker whose load was
# superseded (a new load) or cancelled (unload, incl. an arbiter eviction)
# neither commits its pipeline nor stamps progress onto the current load.
self._load_token = 0
# Set by unload() to abort an in-flight download (which runs without the
# lock, like the chat backend), so an eviction/unload can preempt a slow
# load instead of blocking on the lock for the whole download.
self._cancel_event = threading.Event()
# The callback mutates this and generate_progress() reads it, both without
# the lock (generate holds it for the whole call), so polling stays live.
self._gen: Optional[_GenState] = None
@property
def is_loaded(self) -> bool:
return self._state is not None
def _pick_device_and_dtype(self) -> tuple[str, Any]:
import torch
if torch.cuda.is_available():
# BF16 needs Ampere+ (compute capability >= 8); pre-Ampere cards
# (Turing/Volta/Pascal) only emulate it, so use FP16 there. (Checked by
# capability, not torch.cuda.is_bf16_supported(), which returns True via
# emulation on those cards and would still pick BF16.)
dtype = torch.bfloat16 if torch.cuda.get_device_capability()[0] >= 8 else torch.float16
return "cuda", dtype
# Intel XPU enables the Images page (CHAT_ONLY=False in hardware.py), so
# without this branch an Arc user would silently run diffusion on CPU.
if hasattr(torch, "xpu") and torch.xpu.is_available():
return "xpu", torch.bfloat16
mps = getattr(torch.backends, "mps", None)
if mps is not None and mps.is_available():
return "mps", torch.float16
return "cpu", torch.float32
def _resolve_gguf_path(self, repo_id: str, gguf_filename: str, hf_token: Optional[str]) -> str:
local_root = Path(repo_id).expanduser()
if local_root.exists():
return str(resolve_local_gguf_child(local_root, gguf_filename))
from huggingface_hub import hf_hub_download
return hf_hub_download(repo_id, gguf_filename, token = hf_token)
def _prefetch_files(
self,
repo_id: str,
gguf_filename: Optional[str],
base: str,
base_files: list[str],
hf_token: Optional[str],
) -> None:
"""Pre-download the GGUF + the given ``base_files`` into the HF cache,
WITHOUT the lock and honoring ``_cancel_event``, so load_pipeline's
from_single_file / from_pretrained hit the cache and the heavy download can
be preempted by an unload/eviction. Raises ``RuntimeError("Cancelled")``."""
from utils.hf_xet_fallback import hf_hub_download_with_xet_fallback
# GGUF transformer (hub repos only; a local path is already on disk).
if gguf_filename and not Path(repo_id).expanduser().exists():
hf_hub_download_with_xet_fallback(
repo_id, gguf_filename, hf_token, cancel_event = self._cancel_event
)
# Base repo (VAE / text-encoder / scheduler); list comes from the estimate.
for rfilename in base_files:
if self._cancel_event.is_set():
raise RuntimeError("Cancelled")
hf_hub_download_with_xet_fallback(
base, rfilename, hf_token, cancel_event = self._cancel_event
)
# ── Background load + progress ─────────────────────────────────────────
def validate_load(
self,
repo_id: str,
*,
gguf_filename: Optional[str] = None,
family_override: Optional[str] = None,
) -> DiffusionFamily:
"""Cheap, pure-synchronous pre-flight: returns the resolved family or
raises ValueError. No I/O, so a caller can reject a bad request before
taking the GPU (begin_load and load_pipeline re-check on the load path)."""
if not gguf_filename:
raise ValueError(
"gguf_filename is required: this backend loads single-file GGUF checkpoints only."
)
fam = detect_family(repo_id, family_override)
if fam is None:
raise ValueError(
f"'{repo_id}' isn't a supported image-generation model. "
f"Supported: Z-Image, Qwen-Image, FLUX.1, FLUX.2-klein."
)
return fam
def begin_load(
self,
repo_id: str,
*,
gguf_filename: Optional[str] = None,
base_repo: Optional[str] = None,
family_override: Optional[str] = None,
hf_token: Optional[str] = None,
cpu_offload: bool = False,
) -> dict[str, Any]:
"""Validate, then run the (slow) load on a daemon thread. Returns at once."""
fam = self.validate_load(
repo_id, gguf_filename = gguf_filename, family_override = family_override
)
with self._lock:
# Allow starting over a previously-failed load, but not over a live one.
if self._loading is not None and self._loading.error is None:
raise RuntimeError("A diffusion load is already in progress.")
self._load_token += 1
token = self._load_token
# Best-effort download preemption only; the token (not this event) is
# the real guard that a superseded worker can't commit its pipeline.
self._cancel_event.clear()
# Seed with the family fallback; the worker resolves the real base
# (a network lookup) and updates this, so begin_load never blocks.
self._loading = _LoadingState(repo_id = repo_id, base_repo = fam.base_repo)
threading.Thread(
target = self._run_load,
kwargs = dict(
repo_id = repo_id,
gguf_filename = gguf_filename,
base_repo = base_repo,
family_override = family_override,
hf_token = hf_token,
cpu_offload = cpu_offload,
_load_token = token,
),
daemon = True,
).start()
return self.status()
def _run_load(self, **kwargs: Any) -> None:
token = kwargs.get("_load_token")
try:
# Resolve the base repo and estimate sizes on this thread (both network
# calls) so begin_load returns instantly; the bar shows raw bytes until
# the total lands. This is the only writer of _loading's fields here.
fam = detect_family(kwargs["repo_id"], kwargs.get("family_override"))
base = _resolve_base_repo(
kwargs["repo_id"], kwargs.get("base_repo"), fam, kwargs.get("hf_token")
)
kwargs["base_repo"] = base
expected, base_files = self._estimate_download_bytes(
kwargs["repo_id"], kwargs.get("gguf_filename"), base, kwargs.get("hf_token")
)
with self._lock:
# Stamp progress only if this load is still current; a superseding
# load (or unload) has its own token and its own _LoadingState.
if self._load_token == token and self._loading is not None:
self._loading.base_repo = base
self._loading.expected_bytes = expected
# Download outside the lock so unload()/an eviction can preempt the
# multi-GB pull; load_pipeline below then assembles from the cache.
self._prefetch_files(
kwargs["repo_id"],
kwargs.get("gguf_filename"),
base,
base_files,
kwargs.get("hf_token"),
)
self.load_pipeline(**kwargs)
with self._lock:
# Only clear the marker if this load is still the current one; a
# newer begin_load (or an unload) has its own token.
if self._load_token == token:
self._loading = None
except Exception as exc: # noqa: BLE001 — surfaced to the client via load_progress
# A cancelled/superseded load raised below; don't log it as a failure
# or stamp its error onto whatever load is current now.
if self._load_token != token:
return
logger.error("diffusion.load_failed: %s", exc)
# Redact native paths: this error is surfaced verbatim via the
# load-progress poll, and Studio can run as a shared server.
from utils.native_path_leases import redact_native_paths
with self._lock:
if self._load_token == token and self._loading is not None:
self._loading.error = redact_native_paths(str(exc))
def load_progress(self) -> dict[str, Any]:
"""Phase + downloaded/total bytes for the in-flight load (cache-scan based)."""
loading = self._loading
if loading is not None and loading.error:
return _progress("error", error = loading.error)
if loading is None:
return _progress("ready" if self._state is not None else None)
downloaded = self._cache_bytes(loading.repo_id) + self._cache_bytes(loading.base_repo)
expected = loading.expected_bytes
# Downloads done but pipeline still dequantising / moving to GPU. The cache
# scan can slightly exceed the estimate (extra cached quants, blob padding),
# so clamp the reported bytes/fraction so the bar never overshoots 100%.
if expected > 0 and downloaded >= expected * 0.999:
return _progress("finalizing", min(downloaded, expected), expected, 1.0)
fraction = min(downloaded / expected, 1.0) if expected > 0 else 0.0
return _progress("downloading", downloaded, expected, fraction)
@staticmethod
def _estimate_download_bytes(
repo_id: str, gguf_filename: Optional[str], base_repo: str, hf_token: Optional[str]
) -> tuple[int, list[str]]:
"""Total download size for the progress bar, plus the base-repo files to
fetch (the prefetch reuses this list, so the base is listed only once)."""
from huggingface_hub import HfApi
api = HfApi()
total = 0
base_files: list[str] = []
try:
if gguf_filename:
info = api.model_info(repo_id, files_metadata = True, token = hf_token)
total += sum(s.size or 0 for s in info.siblings if s.rfilename == gguf_filename)
base_info = api.model_info(base_repo, files_metadata = True, token = hf_token)
for s in base_info.siblings:
if _base_file_downloaded(s.rfilename):
base_files.append(s.rfilename)
total += s.size or 0
except Exception as exc: # noqa: BLE001 — estimate is best-effort
logger.warning("diffusion.size_estimate_failed: %s", exc)
return total, base_files
@staticmethod
def _cache_bytes(repo_id: str) -> int:
from huggingface_hub import constants
blobs = Path(constants.HF_HUB_CACHE) / f"models--{repo_id.replace('/', '--')}" / "blobs"
total = 0
try:
for entry in blobs.iterdir():
try:
total += entry.stat().st_size
except OSError:
continue # broken symlink / unreadable
except OSError:
return 0 # repo not in cache yet
return total
# ── Synchronous load / generate / unload ───────────────────────────────
def load_pipeline(
self,
repo_id: str,
*,
gguf_filename: Optional[str] = None,
base_repo: Optional[str] = None,
family_override: Optional[str] = None,
hf_token: Optional[str] = None,
cpu_offload: bool = False,
_load_token: Optional[int] = None,
) -> dict[str, Any]:
import diffusers
fam = self.validate_load(
repo_id, gguf_filename = gguf_filename, family_override = family_override
)
base = _resolve_base_repo(repo_id, base_repo, fam, hf_token)
device, dtype = self._pick_device_and_dtype()
with self._lock:
# Bail before the (slow, VRAM-heavy) build if an unload/eviction or a
# newer load superseded this one while we were resolving/downloading —
# otherwise an evicted load would resurrect a pipeline into VRAM.
if _load_token is not None and _load_token != self._load_token:
raise RuntimeError("Diffusion load was cancelled.")
# Free the old pipeline before allocating the new one so two
# checkpoints never sit in VRAM at once.
self._unload_locked()
# Dequantise the GGUF transformer on-device; the VAE / text-encoder /
# scheduler come from the base diffusers repo (the GGUF is transformer-only).
gguf_path = self._resolve_gguf_path(repo_id, gguf_filename, hf_token)
transformer_cls = getattr(diffusers, fam.transformer_class)
transformer = transformer_cls.from_single_file(
gguf_path,
quantization_config = diffusers.GGUFQuantizationConfig(compute_dtype = dtype),
torch_dtype = dtype,
config = base,
subfolder = "transformer",
# Forward the token: the config is fetched from the (possibly gated)
# base repo before from_pretrained gets a chance to authenticate.
token = hf_token,
)
pipe_kwargs: dict[str, Any] = {"torch_dtype": dtype, "transformer": transformer}
if hf_token:
pipe_kwargs["token"] = hf_token
pipeline_cls = getattr(diffusers, fam.pipeline_class)
pipe = pipeline_cls.from_pretrained(base, **pipe_kwargs)
use_offload = bool(cpu_offload and device == "cuda")
if use_offload:
pipe.enable_model_cpu_offload()
else:
pipe.to(device)
self._state = _LoadState(
pipe = pipe,
family = fam,
repo_id = repo_id,
base_repo = base,
device = device,
dtype = str(dtype).replace("torch.", ""),
cpu_offload = use_offload,
)
logger.info("diffusion.loaded: repo=%s base=%s device=%s", repo_id, base, device)
return self.status()
def generate(
self,
*,
prompt: str,
negative_prompt: Optional[str] = None,
width: int = 1024,
height: int = 1024,
# Fallbacks for a caller that passes nothing; the route always sends the
# per-model values the UI seeds (few steps / no CFG for distilled models,
# more steps / real CFG for full ones).
steps: int = 9,
guidance: float = 0.0,
seed: Optional[int] = None,
batch_size: int = 1,
) -> dict[str, Any]:
import torch
with self._lock:
state = self._state
if state is None:
raise RuntimeError("No diffusion model is loaded.")
generator = torch.Generator(device = state.device)
if seed is None:
# Draw a fresh random seed but keep it within JS's safe-integer
# range (< 2**53), so the reported seed round-trips through JSON
# and actually reproduces the image (a raw 64-bit seed would lose
# precision in the browser and the recipe couldn't be replayed).
seed = generator.seed() & ((1 << 53) - 1)
else:
seed = int(seed)
generator.manual_seed(seed)
kwargs: dict[str, Any] = {
"prompt": prompt,
"width": width,
"height": height,
"num_inference_steps": steps,
# Most pipelines take guidance via "guidance_scale"; Qwen-Image
# uses "true_cfg_scale" (its distilled guidance is off).
state.family.cfg_kwarg: guidance,
"generator": generator,
# Generate the whole batch in one forward pass (VRAM-heavy). All
# share this call's seed, drawn sequentially from one generator.
"num_images_per_prompt": batch_size,
}
# Pipelines vary in which kwargs they accept (a distilled pipeline may
# take neither a negative prompt nor a step callback), so only pass
# those where the signature has them.
call_params = inspect.signature(state.pipe.__call__).parameters
if negative_prompt and "negative_prompt" in call_params:
kwargs["negative_prompt"] = negative_prompt
gen = _GenState(total_steps = steps)
def _on_step(pipe, step_index, timestep, callback_kwargs):
now = time.time()
gen.step = step_index + 1
if gen.first_step_at == 0.0:
gen.first_step_at = now
gen.eta_seconds = _estimate_eta(gen.total_steps, gen.step, gen.first_step_at, now)
return callback_kwargs
if "callback_on_step_end" in call_params:
kwargs["callback_on_step_end"] = _on_step
self._gen = gen
try:
images = state.pipe(**kwargs).images
finally:
self._gen = None
# Return the PIL images (not yet encoded): the route embeds each
# image's recipe and persists it via the gallery.
return {"images": list(images), "seed": int(seed), "repo_id": state.repo_id}
def generate_progress(self) -> dict[str, Any]:
"""Live per-step progress for an in-flight generation (lock-free read)."""
gen = self._gen
if gen is None or gen.total_steps <= 0:
return {
"active": False,
"step": 0,
"total_steps": 0,
"fraction": 0.0,
"eta_seconds": None,
}
return {
"active": True,
"step": gen.step,
"total_steps": gen.total_steps,
"fraction": gen.step / gen.total_steps, # step is 1..total, never over 1.0
"eta_seconds": gen.eta_seconds,
}
def unload(self) -> dict[str, Any]:
# Abort an in-flight download (it runs without the lock and checks this),
# so unload/an eviction returns promptly instead of waiting it out.
self._cancel_event.set()
with self._lock:
self._unload_locked()
# Cancel any in-flight load (its worker checks this token before
# committing) and drop the marker so the next load starts clean.
self._load_token += 1
self._loading = None
return self.status()
def _unload_locked(self) -> None:
state = self._state
if state is None:
return
self._state = None
del state
clear_gpu_cache()
def status(self) -> dict[str, Any]:
state = self._state
if state is None:
return {
"loaded": False,
"repo_id": None,
"family": None,
"base_repo": None,
"device": None,
"dtype": None,
"cpu_offload": False,
}
return {
"loaded": True,
"repo_id": state.repo_id,
"family": state.family.name,
"base_repo": state.base_repo,
"device": state.device,
"dtype": state.dtype,
"cpu_offload": state.cpu_offload,
}
def _resolve_base_repo(
repo_id: str, base_repo: Optional[str], fam: DiffusionFamily, hf_token: Optional[str]
) -> str:
"""The companion diffusers repo: caller's base, else the GGUF repo's own
``base_model`` tag, else the family fallback. Shared by both load paths so a
direct ``load_pipeline`` call resolves the variant base the same way."""
return resolve_base_repo(fam, (base_repo or "").strip() or _hf_base_model(repo_id, hf_token))
def _hf_base_model(repo_id: str, hf_token: Optional[str]) -> Optional[str]:
"""The diffusers base repo from a GGUF repo's ``base_model`` tag, or None.
Lets one family entry cover every variant (Turbo/full, schnell/dev, the
2512 Qwen revision). Skipped for local paths; None on any lookup failure.
"""
if Path(repo_id).expanduser().exists():
return None
try:
from huggingface_hub import HfApi
meta = HfApi().model_info(repo_id, token = hf_token).cardData or {}
except Exception: # noqa: BLE001 — best-effort; fall back to the family default
return None
base = meta.get("base_model")
if isinstance(base, list):
base = base[0] if base else None
return base if isinstance(base, str) and base.strip() else None
def _base_file_downloaded(rfilename: str) -> bool:
"""True for base-repo files ``from_pretrained`` actually fetches.
The transformer is supplied by the GGUF, and repo docs (``assets/``, the
top-level README/PDF/images) are never downloaded — counting them would peg
the progress estimate above what lands on disk, so the bar would sit short of
100% for the whole pipeline-load phase instead of advancing to "finalizing".
"""
if rfilename.startswith("transformer/"):
return False
if "/" not in rfilename: # top-level: only the pipeline manifest is fetched
return rfilename == "model_index.json"
return not rfilename.startswith("assets/")
def _progress(
phase: Optional[str],
bytes_downloaded: int = 0,
bytes_total: int = 0,
fraction: float = 0.0,
*,
error: Optional[str] = None,
) -> dict[str, Any]:
return {
"phase": phase,
"bytes_downloaded": bytes_downloaded,
"bytes_total": bytes_total,
"fraction": fraction,
"error": error,
}
_diffusion_backend: Optional[DiffusionBackend] = None
def get_diffusion_backend() -> DiffusionBackend:
global _diffusion_backend
if _diffusion_backend is None:
_diffusion_backend = DiffusionBackend()
return _diffusion_backend