Compare commits
47 Commits
b750d26d80
...
feat/herme
| Author | SHA1 | Date | |
|---|---|---|---|
| e607982b21 | |||
| 4627199217 | |||
| 3611d645db | |||
| 99ac8b5dcd | |||
| b517a65393 | |||
| 61a59dfac1 | |||
| 144678354f | |||
| 34b1cb83a0 | |||
| f8c2f864de | |||
| 46241d10c5 | |||
| 6b82a26c25 | |||
| 9cc7edfb39 | |||
| f31ae59717 | |||
| bf2f17c5e2 | |||
| 208bfd4612 | |||
| 32df546550 | |||
| 0f7b22c19b | |||
| fc777e2de2 | |||
| d52f18b0fa | |||
| 0d87fb2556 | |||
| f6bc2b07a7 | |||
| aa6bbe87bf | |||
| 0c612d9731 | |||
| 5b210fe624 | |||
| d8b77c97c3 | |||
| a3d0fa0072 | |||
| 956d76f14d | |||
| c6d2f5918f | |||
| f023dc1ee4 | |||
| d34a4d3647 | |||
| ef58155897 | |||
| 0580603f27 | |||
| 93c01fa314 | |||
| 6e540635bf | |||
|
|
b89be7b8f4 | ||
|
|
748b5037b9 | ||
|
|
3f80744ebd | ||
|
|
6f17743667 | ||
|
|
98216d2872 | ||
|
|
a40e347dfa | ||
|
|
cfa2a898c3 | ||
|
|
0609720b33 | ||
|
|
d97f1cb1e5 | ||
|
|
1a1cfec80a | ||
|
|
90e227bc4e | ||
|
|
8e9a75fe5c | ||
|
|
3016d0da2c |
31
.gitea/workflows/build-hermes.yml
Normal file
31
.gitea/workflows/build-hermes.yml
Normal file
@@ -0,0 +1,31 @@
|
||||
name: Build Hermes agent
|
||||
on:
|
||||
pull_request:
|
||||
branches: [ master ]
|
||||
paths:
|
||||
- 'ai/hermes/**'
|
||||
- 'ai/compose.yml'
|
||||
push:
|
||||
branches: [ master ]
|
||||
paths:
|
||||
- 'ai/hermes/**'
|
||||
- 'ai/compose.yml'
|
||||
|
||||
jobs:
|
||||
build:
|
||||
runs-on: ubuntu-latest
|
||||
steps:
|
||||
- name: Checkout
|
||||
run: |
|
||||
git clone -b "${{ github.head_ref || github.ref_name }}" \
|
||||
https://gitea:${{ secrets.GITHUB_TOKEN }}@code.lazyworkhorse.net/gortium/compose.git .
|
||||
git log --oneline -3
|
||||
|
||||
- name: Build hermes image
|
||||
run: |
|
||||
cd ai
|
||||
docker compose build hermes 2>&1
|
||||
|
||||
- name: Verify image
|
||||
run: |
|
||||
docker run --rm ai-hermes /opt/hermes/.venv/bin/python --version 2>&1
|
||||
31
.gitea/workflows/build-ollama.yml
Normal file
31
.gitea/workflows/build-ollama.yml
Normal file
@@ -0,0 +1,31 @@
|
||||
name: Build ollama (gfx906)
|
||||
on:
|
||||
pull_request:
|
||||
branches: [ master ]
|
||||
paths:
|
||||
- 'ai/ollama/**'
|
||||
- 'ai/compose.yml'
|
||||
push:
|
||||
branches: [ master ]
|
||||
paths:
|
||||
- 'ai/ollama/**'
|
||||
- 'ai/compose.yml'
|
||||
|
||||
jobs:
|
||||
build:
|
||||
runs-on: ubuntu-latest
|
||||
steps:
|
||||
- name: Checkout
|
||||
run: |
|
||||
git clone -b "${{ github.head_ref || github.ref_name }}" \
|
||||
https://gitea:${{ secrets.GITHUB_TOKEN }}@code.lazyworkhorse.net/gortium/compose.git .
|
||||
git log --oneline -3
|
||||
|
||||
- name: Build ollama image
|
||||
run: |
|
||||
cd ai
|
||||
docker compose build ollama --no-cache 2>&1
|
||||
|
||||
- name: Verify version
|
||||
run: |
|
||||
docker run --rm ollama/ollama:rocm-gfx906 ollama --version 2>&1
|
||||
@@ -1,67 +0,0 @@
|
||||
# 1. On récupère la version la plus récente d'UV
|
||||
FROM ghcr.io/astral-sh/uv:latest AS uv_source
|
||||
|
||||
# 2. Image de base stable
|
||||
FROM debian:stable-slim
|
||||
|
||||
# Disable Python stdout buffering to ensure logs are printed immediately
|
||||
ENV PYTHONUNBUFFERED=1
|
||||
|
||||
# Install system dependencies in one layer, clear APT cache
|
||||
# tini reaps orphaned zombie processes (MCP stdio subprocesses, git, bun, etc.)
|
||||
RUN apt-get update && \
|
||||
apt-get install -y --no-install-recommends \
|
||||
build-essential python3 ripgrep ffmpeg gcc python3-dev libffi-dev procps git openssh-client docker-cli tini \
|
||||
curl poppler-utils imagemagick \
|
||||
texlive-latex-base texlive-latex-extra texlive-fonts-recommended texlive-xetex texlive-science \
|
||||
qemu-user-static binfmt-support qemu-user-binfmt \
|
||||
emacs-nox \
|
||||
libportaudio2 \
|
||||
ca-certificates && \
|
||||
rm -rf /var/lib/apt/lists/*
|
||||
|
||||
# Création de l'utilisateur 'hermes' directement avec les bons accès
|
||||
RUN useradd -u 10000 -m -d /opt/data hermes
|
||||
|
||||
# Copie d'uv (dernière version)
|
||||
COPY --chmod=0755 --from=uv_source /uv /usr/local/bin/
|
||||
|
||||
WORKDIR /opt/hermes
|
||||
|
||||
# On donne la propriété du dossier de travail à l'utilisateur hermes
|
||||
RUN chown hermes:hermes /opt/hermes
|
||||
|
||||
# ---------- Hermes venv ----------
|
||||
# Passer immédiatement sous l'utilisateur hermes pour tout le reste du build
|
||||
USER hermes
|
||||
|
||||
# ---------- Source code ----------
|
||||
# On copie tout le projet d'un coup sans assumer la présence de fichiers de lock spécifiques
|
||||
COPY --chown=hermes:hermes . .
|
||||
|
||||
# ---------- Python virtualenv avec Piper TTS ----------
|
||||
RUN uv venv && \
|
||||
uv pip install --no-cache-dir piper-tts sounddevice numpy faster-whisper
|
||||
|
||||
# ---------- Télécharger la voix Piper Norman ----------
|
||||
RUN mkdir -p /opt/hermes/.venv/share/piper/voices && \
|
||||
/opt/hermes/.venv/bin/python3 /dev/stdin << 'PYEOF' && rm -f /tmp/dl_voice.py
|
||||
import urllib.request
|
||||
base = '/opt/hermes/.venv/share/piper/voices'
|
||||
url = 'https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/norman/medium/en_US-norman-medium.onnx'
|
||||
urllib.request.urlretrieve(url, base + '/en_US-norman-medium.onnx')
|
||||
urllib.request.urlretrieve(url + '.json', base + '/en_US-norman-medium.onnx.json')
|
||||
PYEOF
|
||||
|
||||
# ---------- Patch tts_tool.py: replace Edge TTS fallback with Piper ----------
|
||||
COPY patch_tts_tool.py /tmp/patch_tts_tool.py
|
||||
RUN /opt/hermes/.venv/bin/python3 /tmp/patch_tts_tool.py && rm /tmp/patch_tts_tool.py
|
||||
|
||||
# ---------- Runtime ----------
|
||||
ENV HERMES_HOME=/opt/data
|
||||
ENV PATH="/opt/data/.local/bin:${PATH}"
|
||||
|
||||
VOLUME [ "/opt/data" ]
|
||||
|
||||
# Le conteneur tourne de manière ultra-sécurisée sous l'utilisateur hermes dès le départ
|
||||
ENTRYPOINT [ "/usr/bin/tini", "-g", "--", "/opt/hermes/docker/entrypoint.sh" ]
|
||||
@@ -26,8 +26,14 @@ services:
|
||||
# - "traefik.http.routers.webui-https.tls.certresolver=njalla"
|
||||
|
||||
hermes:
|
||||
build: ./
|
||||
build:
|
||||
context: ./hermes
|
||||
ssh:
|
||||
- default
|
||||
container_name: hermes
|
||||
entrypoint: ["/bin/bash", "-c",
|
||||
"bash /opt/data/hermes-tools/install.sh && exec /usr/bin/tini -g -- /opt/hermes/docker/entrypoint.sh \"$@\"",
|
||||
"hermes-entrypoint"]
|
||||
restart: always
|
||||
# Gateway run enables the internal API server on port 8642
|
||||
command: gateway run
|
||||
@@ -45,6 +51,7 @@ services:
|
||||
- HIP_VISIBLE_DEVICES=0,1
|
||||
- ROCR_VISIBLE_DEVICES=0,1
|
||||
- HSA_ENABLE_SDMA=0
|
||||
- TZ=America/Montreal
|
||||
volumes:
|
||||
- /mnt/HoardingCow_docker_data/Hermes/data:/opt/data
|
||||
devices:
|
||||
@@ -57,9 +64,11 @@ services:
|
||||
- ai_backend
|
||||
|
||||
ollama:
|
||||
image: ollama/ollama:latest
|
||||
build:
|
||||
context: ./ollama
|
||||
dockerfile: Dockerfile
|
||||
image: ollama/ollama:rocm-gfx906
|
||||
container_name: ollama
|
||||
privileged: true
|
||||
tty: true
|
||||
restart: always
|
||||
ports:
|
||||
@@ -77,7 +86,7 @@ services:
|
||||
- HSA_ENABLE_SDMA=0
|
||||
- OLLAMA_HOST=0.0.0.0
|
||||
- OLLAMA_DEBUG=1
|
||||
- OLLAMA_FLASH_ATTENTION=0
|
||||
- OLLAMA_FLASH_ATTENTION=1
|
||||
- OLLAMA_NUM_PARALLEL=2
|
||||
devices:
|
||||
# Map the render nodes and KFD for ROCm to work inside the container
|
||||
|
||||
75
ai/hermes/Dockerfile
Normal file
75
ai/hermes/Dockerfile
Normal file
@@ -0,0 +1,75 @@
|
||||
# syntax=docker/dockerfile:1
|
||||
# Hermes Agent -- custom fork build
|
||||
# Builds on top of official image + overlays our forked source from Gitea.
|
||||
# Requires Docker BuildKit. Pass SSH agent for git clone:
|
||||
# docker compose build hermes
|
||||
# Or manually:
|
||||
# DOCKER_BUILDKIT=1 docker build --ssh default -t hermes-agent:custom .
|
||||
|
||||
# ---------- Base: official Hermes image (system deps, npm, uv, Playwright) ----------
|
||||
FROM nousresearch/hermes-agent:latest
|
||||
|
||||
# ---------- Overlay our forked source ----------
|
||||
# Uses SSH agent forwarding from the build host (no key baked into image).
|
||||
# --exclude node_modules/.venv keeps the base image's pre-built layers intact.
|
||||
# Only the Python source, web UI source, and config change.
|
||||
RUN --mount=type=ssh \
|
||||
mkdir -p /root/.ssh && \
|
||||
ssh-keyscan -p 2222 code.lazyworkhorse.net >> /root/.ssh/known_hosts 2>/dev/null && \
|
||||
cd /tmp && \
|
||||
GIT_SSH_COMMAND='ssh -p 2222 -o StrictHostKeyChecking=no' \
|
||||
git clone --depth 1 --branch main \
|
||||
git@code.lazyworkhorse.net:gortium/hermes-agent.git fork && \
|
||||
rsync -a --delete fork/ /opt/hermes/ \
|
||||
--exclude node_modules \
|
||||
--exclude .venv \
|
||||
--exclude .git && \
|
||||
rm -rf /tmp/fork /root/.ssh/
|
||||
|
||||
# ---------- Rebuild web UI ----------
|
||||
# Source files changed; node_modules (from base image) reused.
|
||||
RUN cd /opt/hermes && npm run build
|
||||
|
||||
# ---------- Reinstall Python package (editable) ----------
|
||||
# Picks up source changes from our fork.
|
||||
RUN . /opt/hermes/.venv/bin/activate && \
|
||||
uv pip install --no-cache-dir --no-deps -e /opt/hermes
|
||||
|
||||
# ---------- Extra system deps ----------
|
||||
USER root
|
||||
RUN apt-get update && \
|
||||
apt-get install -y --no-install-recommends \
|
||||
libportaudio2 ca-certificates poppler-utils imagemagick \
|
||||
texlive-latex-base texlive-latex-extra texlive-fonts-recommended \
|
||||
texlive-xetex texlive-science \
|
||||
qemu-user-static binfmt-support emacs-nox && \
|
||||
rm -rf /var/lib/apt/lists/*
|
||||
|
||||
# ---------- UV ----------
|
||||
COPY --chmod=0755 --from=ghcr.io/astral-sh/uv:latest /uv /usr/local/bin/
|
||||
|
||||
# ---------- Piper TTS ----------
|
||||
RUN . /opt/hermes/.venv/bin/activate && \
|
||||
uv pip install --no-cache-dir piper-tts sounddevice numpy && \
|
||||
mkdir -p /opt/hermes/.venv/share/piper/voices
|
||||
|
||||
RUN /opt/hermes/.venv/bin/python3 /dev/stdin << 'PYEOF'
|
||||
import urllib.request
|
||||
base = '/opt/hermes/.venv/share/piper/voices'
|
||||
url = 'https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/ryan/high/en_US-ryan-high.onnx'
|
||||
urllib.request.urlretrieve(url, base + '/en_US-ryan-high.onnx')
|
||||
urllib.request.urlretrieve(url + '.json', base + '/en_US-ryan-high.onnx.json')
|
||||
PYEOF
|
||||
|
||||
# ---------- Runtime ----------
|
||||
USER hermes
|
||||
ENV HERMES_HOME=/opt/data
|
||||
ENV PATH="/opt/data/.local/bin:${PATH}"
|
||||
# Point browser tool to Playwright's Chromium (already in base image)
|
||||
ENV CHROME_EXECUTABLE=/opt/hermes/.playwright/chromium/chrome-linux/chrome
|
||||
|
||||
# Ensure tools directory and toolsets.py are writable by the hermes runtime user
|
||||
# so custom tools can be injected from the persistent volume at startup.
|
||||
RUN chown -R hermes:hermes /opt/hermes/tools /opt/hermes/toolsets.py
|
||||
|
||||
VOLUME [ "/opt/data" ]
|
||||
181
ai/hermes/patch_tts_tool.py
Normal file
181
ai/hermes/patch_tts_tool.py
Normal file
@@ -0,0 +1,181 @@
|
||||
#!/usr/bin/env python3
|
||||
"""Patch Hermes TTS tool: add Piper TTS provider, remove Edge TTS as default.
|
||||
|
||||
Patches ALL copies of tts_tool.py found (venv site-packages + /opt/hermes/tools/).
|
||||
|
||||
Searches multiple paths for tts_tool.py so it works both at build time
|
||||
(in the image venv) and at runtime (on the mounted data volume).
|
||||
|
||||
Idempotent: if already patched, does nothing.
|
||||
"""
|
||||
|
||||
import sys
|
||||
import os
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Search for all copies of tts_tool.py
|
||||
# ---------------------------------------------------------------------------
|
||||
CANDIDATE_PATHS = [
|
||||
"/opt/hermes/.venv/lib/python3.13/site-packages/tools/tts_tool.py",
|
||||
"/opt/hermes/tools/tts_tool.py",
|
||||
]
|
||||
|
||||
found_paths = []
|
||||
|
||||
for p in CANDIDATE_PATHS:
|
||||
if os.path.exists(p):
|
||||
found_paths.append(p)
|
||||
print(f"Found tts_tool.py at: {p}")
|
||||
|
||||
# Also try to find via Python import
|
||||
import subprocess
|
||||
try:
|
||||
result = subprocess.run(
|
||||
[sys.executable, "-c", "import tools.tts_tool; print(tools.tts_tool.__file__)"],
|
||||
capture_output=True, text=True, timeout=5
|
||||
)
|
||||
if result.returncode == 0:
|
||||
p = result.stdout.strip()
|
||||
if os.path.exists(p) and p not in found_paths:
|
||||
found_paths.append(p)
|
||||
print(f"Found tts_tool.py via import at: {p}")
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
if not found_paths:
|
||||
print("WARNING: tts_tool.py not found anywhere. Patching deferred to runtime.")
|
||||
print(f"Searched: {CANDIDATE_PATHS}")
|
||||
sys.exit(0)
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Old else block: the Edge TTS default fallback to replace
|
||||
# ---------------------------------------------------------------------------
|
||||
old_else = ''' else:
|
||||
# Default: Edge TTS (free), with NeuTTS as local fallback
|
||||
edge_available = True
|
||||
try:
|
||||
_import_edge_tts()
|
||||
except ImportError:
|
||||
edge_available = False
|
||||
|
||||
if edge_available:
|
||||
logger.info("Generating speech with Edge TTS...")
|
||||
try:
|
||||
import concurrent.futures
|
||||
with concurrent.futures.ThreadPoolExecutor(max_workers=1) as pool:
|
||||
pool.submit(
|
||||
lambda: asyncio.run(_generate_edge_tts(text, file_str, tts_config))
|
||||
).result(timeout=60)
|
||||
except RuntimeError:
|
||||
asyncio.run(_generate_edge_tts(text, file_str, tts_config))
|
||||
elif _check_neutts_available():
|
||||
logger.info("Edge TTS not available, falling back to NeuTTS (local)...")
|
||||
provider = "neutts"
|
||||
_generate_neutts(text, file_str, tts_config)
|
||||
else:
|
||||
return json.dumps({
|
||||
"success": False,
|
||||
"error": "No TTS provider available. Install edge-tts (pip install edge-tts) "
|
||||
"or set up NeuTTS for local synthesis."
|
||||
}, ensure_ascii=False)'''
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# New block: elif provider == "piper" + else: fallback with Piper only
|
||||
# ---------------------------------------------------------------------------
|
||||
new_block = ''' elif provider == "piper":
|
||||
# Piper TTS (local, CPU, no cloud, no Microsoft)
|
||||
piper_binary = "/opt/hermes/.venv/bin/piper"
|
||||
piper_config = tts_config.get("piper", {})
|
||||
voice = piper_config.get("voice", "en_US-lessac-medium")
|
||||
model_dir = piper_config.get("model_dir", "/opt/hermes/.venv/share/piper/voices")
|
||||
model_path = os.path.join(model_dir, f"{voice}.onnx")
|
||||
if not os.path.exists(model_path):
|
||||
return json.dumps({
|
||||
"success": False,
|
||||
"error": "Piper TTS voice model not found. "
|
||||
"Install Piper TTS and download a voice model."
|
||||
}, ensure_ascii=False)
|
||||
logger.info("Generating speech with Piper TTS (local, CPU)...")
|
||||
import subprocess as _sp
|
||||
cmd = [piper_binary, "--model", model_path, "--output-raw"]
|
||||
try:
|
||||
proc = _sp.Popen(cmd, stdin=_sp.PIPE, stdout=_sp.PIPE, stderr=_sp.PIPE)
|
||||
raw_audio, stderr = proc.communicate(input=text.encode(), timeout=60)
|
||||
if proc.returncode != 0:
|
||||
raise RuntimeError(f"Piper TTS failed: {stderr.decode()[:200]}")
|
||||
ffmpeg_cmd = ["ffmpeg", "-f", "s16le", "-ar", "22050", "-ac", "1", "-i", "-", "-y", file_str]
|
||||
_sp.run(ffmpeg_cmd, input=raw_audio, capture_output=True, timeout=30)
|
||||
except Exception as e:
|
||||
return json.dumps({
|
||||
"success": False,
|
||||
"error": f"Piper TTS failed: {e}"
|
||||
}, ensure_ascii=False)
|
||||
|
||||
else:
|
||||
# Default: Piper TTS (local, CPU, no cloud, no Microsoft)
|
||||
piper_binary = "/opt/hermes/.venv/bin/piper"
|
||||
piper_config = tts_config.get("piper", {})
|
||||
voice = piper_config.get("voice", "en_US-lessac-medium")
|
||||
model_dir = piper_config.get("model_dir", "/opt/hermes/.venv/share/piper/voices")
|
||||
model_path = os.path.join(model_dir, f"{voice}.onnx")
|
||||
if os.path.exists(model_path) and os.path.exists(piper_binary):
|
||||
logger.info("Generating speech with Piper TTS (local, CPU)...")
|
||||
import subprocess as _sp
|
||||
cmd = [piper_binary, "--model", model_path, "--output-raw"]
|
||||
try:
|
||||
proc = _sp.Popen(cmd, stdin=_sp.PIPE, stdout=_sp.PIPE, stderr=_sp.PIPE)
|
||||
raw_audio, stderr = proc.communicate(input=text.encode(), timeout=60)
|
||||
if proc.returncode != 0:
|
||||
raise RuntimeError(stderr.decode()[:200])
|
||||
ffmpeg_cmd = ["ffmpeg", "-f", "s16le", "-ar", "22050", "-ac", "1", "-i", "-", "-y", file_str]
|
||||
_sp.run(ffmpeg_cmd, input=raw_audio, capture_output=True, timeout=30)
|
||||
except Exception:
|
||||
pass
|
||||
else:
|
||||
return json.dumps({
|
||||
"success": False,
|
||||
"error": "Piper TTS not available. Install piper-tts and download a voice model."
|
||||
}, ensure_ascii=False)'''
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Apply the patch to all copies found
|
||||
# ---------------------------------------------------------------------------
|
||||
patched_any = False
|
||||
|
||||
for tts_path in found_paths:
|
||||
with open(tts_path) as f:
|
||||
code = f.read()
|
||||
|
||||
if 'provider == "piper"' in code:
|
||||
print(f"ALREADY PATCHED: {tts_path}")
|
||||
continue
|
||||
|
||||
if old_else in code:
|
||||
code = code.replace(old_else, new_block, 1)
|
||||
with open(tts_path, 'w') as f:
|
||||
f.write(code)
|
||||
print(f"PATCHED: {tts_path}")
|
||||
patched_any = True
|
||||
else:
|
||||
print(f"SKIP {tts_path}: Edge fallback pattern not found")
|
||||
import re
|
||||
for m in re.finditer(r' else:\n # Default:', code):
|
||||
start = max(0, m.start() - 100)
|
||||
end = min(len(code), m.end() + 300)
|
||||
print(f" Found 'else:/# Default:' at position {m.start()}:")
|
||||
print(f" {code[start:end]}")
|
||||
print(" ---")
|
||||
# Don't exit with error — if one copy isn't patchable, try the others
|
||||
|
||||
if not patched_any:
|
||||
all_patched = all(
|
||||
'provider == "piper"' in open(p).read()
|
||||
for p in found_paths
|
||||
)
|
||||
if all_patched:
|
||||
print("All copies already patched.")
|
||||
sys.exit(0)
|
||||
print("WARNING: Could not patch any copy of tts_tool.py")
|
||||
sys.exit(1)
|
||||
|
||||
print("tts_tool.py patched successfully across all copies.")
|
||||
106
ai/ollama/Dockerfile
Normal file
106
ai/ollama/Dockerfile
Normal file
@@ -0,0 +1,106 @@
|
||||
# ollama-gfx906/Dockerfile
|
||||
#
|
||||
# Custom ollama image with ROCm 6.1 + gfx906 (MI50) support.
|
||||
# The official ollama/rocm image ships ROCm 7.2 which dropped gfx906.
|
||||
# This uses v0.23.2's native CMake build system with AMDGPU_TARGETS including gfx906.
|
||||
#
|
||||
# Build: docker build -t ollama/ollama:rocm-gfx906 ai/ollama
|
||||
|
||||
FROM rocm/dev-ubuntu-22.04:6.1.2-complete AS builder
|
||||
|
||||
# Build dependencies (CMake, Ninja, Go)
|
||||
ARG CMAKEVERSION=3.31.2
|
||||
ARG NINJAVERSION=1.12.1
|
||||
ARG GOLANG_VERSION=1.22.0
|
||||
|
||||
RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
|
||||
curl git ccache build-essential pkg-config unzip \
|
||||
&& rm -rf /var/lib/apt/lists/*
|
||||
|
||||
# Install CMake from official binaries
|
||||
RUN curl -fsSL https://github.com/Kitware/CMake/releases/download/v${CMAKEVERSION}/cmake-${CMAKEVERSION}-linux-x86_64.tar.gz \
|
||||
| tar xz -C /usr/local --strip-components 1
|
||||
|
||||
# Install Ninja
|
||||
RUN curl -fsSL -o /tmp/ninja.zip \
|
||||
https://github.com/ninja-build/ninja/releases/download/v${NINJAVERSION}/ninja-linux.zip \
|
||||
&& unzip /tmp/ninja.zip -d /usr/local/bin && rm /tmp/ninja.zip
|
||||
|
||||
# Install Go
|
||||
RUN curl -fsSL https://go.dev/dl/go${GOLANG_VERSION}.linux-amd64.tar.gz \
|
||||
| tar xz -C /usr/local
|
||||
ENV PATH=/usr/local/go/bin:$PATH
|
||||
|
||||
ARG OLLAMA_VERSION=v0.23.2
|
||||
RUN git clone --depth 1 --branch ${OLLAMA_VERSION} https://github.com/ollama/ollama.git /build
|
||||
WORKDIR /build
|
||||
|
||||
# ROCm paths
|
||||
ENV HIP_PATH=/opt/rocm
|
||||
ENV ROCM_PATH=/opt/rocm
|
||||
ENV CMAKE_GENERATOR=Ninja
|
||||
ENV LDFLAGS=-s
|
||||
|
||||
# Step 1: Build CPU backends with GCC (no ROCm preset)
|
||||
# Pre-set CMAKE_HIP_COMPILER="" to prevent check_language(HIP) from
|
||||
# finding a HIP compiler (it searches /opt/rocm even without PATH).
|
||||
# Remove /opt/rocm from PATH to prevent find_program from finding hipcc.
|
||||
RUN mkdir -p build-cpu && \
|
||||
PATH=/usr/local/go/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin \
|
||||
cmake -B build-cpu -DCMAKE_BUILD_TYPE=Release \
|
||||
-DCMAKE_HIP_COMPILER="" \
|
||||
-DCMAKE_INSTALL_PREFIX=/build/dist && \
|
||||
cmake --build build-cpu --target ggml-cpu -- -l $(nproc) && \
|
||||
cmake --install build-cpu --component CPU --strip && \
|
||||
echo "=== CPU install ===" && \
|
||||
(find /build/dist/lib/ollama -type f -o -type l 2>&1 | head -20 || echo "empty")
|
||||
|
||||
# Step 2: Build HIP backend with ROCm preset + gfx906 target only
|
||||
# The ROCm 6 preset enables HIP language detection (enable_language(HIP))
|
||||
# which ensures GPU kernels are properly compiled for gfx906.
|
||||
# OLLAMA_RUNNER_DIR=rocm from the preset, so HIP goes to lib/ollama/rocm/
|
||||
# Need CMAKE_PREFIX_PATH so find_package(hip) finds hip-config.cmake
|
||||
# at /opt/rocm/lib/cmake/hip/hip-config.cmake.
|
||||
RUN mkdir -p build-hip && \
|
||||
cmake -B build-hip \
|
||||
--preset 'ROCm 6' \
|
||||
-DAMDGPU_TARGETS="gfx906:xnack-" \
|
||||
-DCMAKE_BUILD_TYPE=Release \
|
||||
-DCMAKE_PREFIX_PATH="/opt/rocm" && \
|
||||
cmake --build build-hip --target ggml-hip -- -l $(nproc) && \
|
||||
cmake --install build-hip --component HIP --strip && \
|
||||
echo "=== HIP install ===" && \
|
||||
find /build/dist/lib/ollama -type f -o -type l | head -20
|
||||
|
||||
# Step 3: Build Go binary (GCC for CGo linking)
|
||||
ENV CGO_ENABLED=1
|
||||
RUN go build -trimpath -ldflags="-X=github.com/ollama/ollama/version.Version=${OLLAMA_VERSION}" -o /build/dist/ollama .
|
||||
|
||||
# ---------- Runtime image ----------
|
||||
FROM ubuntu:24.04
|
||||
|
||||
RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
|
||||
ca-certificates curl libstdc++6 libgomp1 libvulkan1 libopenblas0 \
|
||||
&& rm -rf /var/lib/apt/lists/*
|
||||
|
||||
# Copy ROCm 6.1 runtime libraries
|
||||
# These are needed at runtime by ggml-hip via LD_LIBRARY_PATH
|
||||
COPY --from=builder /opt/rocm/lib/ /opt/rocm/lib/
|
||||
COPY --from=builder /opt/rocm/share/ /opt/rocm/share/
|
||||
|
||||
# Copy ollama binary + all backends (CPU + HIP)
|
||||
# CPU install: /build/dist/lib/ollama/libggml-*.so
|
||||
# HIP install: /build/dist/lib/ollama/rocm/libggml-hip.so
|
||||
COPY --from=builder /build/dist/ollama /usr/bin/ollama
|
||||
COPY --from=builder /build/dist/lib/ollama/ /usr/lib/ollama/
|
||||
|
||||
RUN ldconfig
|
||||
|
||||
ENV LD_LIBRARY_PATH=/opt/rocm/lib:/usr/lib/ollama/rocm:/usr/lib/ollama
|
||||
ENV HSA_OVERRIDE_GFX_VERSION=9.0.6
|
||||
ENV HCC_AMDGPU_TARGET=gfx906
|
||||
ENV HSA_ENABLE_SDMA=0
|
||||
|
||||
EXPOSE 11434
|
||||
ENTRYPOINT ["/bin/ollama"]
|
||||
CMD ["serve"]
|
||||
@@ -1,96 +0,0 @@
|
||||
#!/usr/bin/env python3
|
||||
"""Patch Hermes TTS tool: remove Edge TTS, replace with Piper as default/fallback."""
|
||||
import sys
|
||||
|
||||
tts_path = '/opt/hermes/.venv/lib/python3.13/site-packages/tools/tts_tool.py'
|
||||
|
||||
with open(tts_path) as f:
|
||||
code = f.read()
|
||||
|
||||
# Replace the Edge fallback with Piper fallback
|
||||
old_edge = ''' else:
|
||||
# Default: Edge TTS (free), with NeuTTS as local fallback
|
||||
edge_available = True
|
||||
try:
|
||||
_import_edge_tts()
|
||||
except ImportError:
|
||||
edge_available = False
|
||||
|
||||
if edge_available:
|
||||
logger.info("Generating speech with Edge TTS...")
|
||||
try:
|
||||
import concurrent.futures
|
||||
with concurrent.futures.ThreadPoolExecutor(max_workers=1) as pool:
|
||||
pool.submit(
|
||||
lambda: asyncio.run(_generate_edge_tts(text, file_str, tts_config))
|
||||
).result(timeout=60)
|
||||
except RuntimeError:
|
||||
asyncio.run(_generate_edge_tts(text, file_str, tts_config))
|
||||
elif _check_neutts_available():
|
||||
logger.info("Edge TTS not available, falling back to NeuTTS (local)...")
|
||||
provider = "neutts"
|
||||
_generate_neutts(text, file_str, tts_config)
|
||||
else:
|
||||
return json.dumps({
|
||||
"success": False,
|
||||
"error": "No TTS provider available. Install edge-tts (pip install edge-tts) "
|
||||
"or set up NeuTTS for local synthesis."
|
||||
}, ensure_ascii=False)'''
|
||||
|
||||
new_piper = ''' else:
|
||||
# Default: Piper TTS (local, CPU, no cloud, no Microsoft)
|
||||
piper_available = False
|
||||
try:
|
||||
piper_binary = "/opt/hermes/.venv/bin/piper"
|
||||
piper_config = tts_config.get("piper", {})
|
||||
voice = piper_config.get("voice", "en_US-lessac-medium")
|
||||
model_dir = piper_config.get("model_dir", "/opt/hermes/.venv/share/piper/voices")
|
||||
model_path = os.path.join(model_dir, f"{voice}.onnx")
|
||||
if os.path.exists(model_path):
|
||||
piper_available = True
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
if piper_available:
|
||||
logger.info("Generating speech with Piper TTS (local, CPU)...")
|
||||
import subprocess
|
||||
piper_binary = "/opt/hermes/.venv/bin/piper"
|
||||
piper_config = tts_config.get("piper", {})
|
||||
voice = piper_config.get("voice", "en_US-lessac-medium")
|
||||
model_dir = piper_config.get("model_dir", "/opt/hermes/.venv/share/piper/voices")
|
||||
model_path = os.path.join(model_dir, f"{voice}.onnx")
|
||||
cmd = [piper_binary, "--model", model_path, "--output-raw"]
|
||||
proc = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
|
||||
raw_audio, stderr = proc.communicate(input=text.encode(), timeout=60)
|
||||
if proc.returncode != 0:
|
||||
raise RuntimeError(f"Piper TTS failed: {stderr.decode()[:200]}")
|
||||
ffmpeg_cmd = ["ffmpeg", "-f", "s16le", "-ar", "22050", "-ac", "1", "-i", "-", "-y", file_str]
|
||||
subprocess.run(ffmpeg_cmd, input=raw_audio, capture_output=True, timeout=30)
|
||||
logger.info("Piper TTS audio saved: %s", file_str)
|
||||
else:
|
||||
return json.dumps({
|
||||
"success": False,
|
||||
"error": "No TTS provider available. Install Piper TTS (pip install piper-tts) "
|
||||
"and download a voice model."
|
||||
}, ensure_ascii=False)'''
|
||||
|
||||
if old_edge in code:
|
||||
code = code.replace(old_edge, new_piper)
|
||||
print("Edge fallback replaced with Piper")
|
||||
else:
|
||||
if 'Default: Piper TTS' in code:
|
||||
print("Piper fallback already present")
|
||||
else:
|
||||
print("ERROR: Could not find Edge fallback in tts_tool.py")
|
||||
# Debug output
|
||||
import re
|
||||
for m in re.finditer(r' else:\n # Default:', code):
|
||||
start = max(0, m.start() - 100)
|
||||
end = min(len(code), m.end() + 200)
|
||||
print(f"Found else/default at position {m.start()}:")
|
||||
print(code[start:end])
|
||||
sys.exit(1)
|
||||
|
||||
with open(tts_path, 'w') as f:
|
||||
f.write(code)
|
||||
print("tts_tool.py patched successfully")
|
||||
@@ -7,8 +7,14 @@ services:
|
||||
- USER_UID=1000
|
||||
- USER_GID=1000
|
||||
- GITEA__server__ROOT_URL=https://code.lazyworkhorse.net
|
||||
- GITEA__actions__ENABLED=true
|
||||
- GITEA__actions__DEFAULT_ACTIONS_URL=off
|
||||
- SSH_PORT=2222
|
||||
- SSH_LISTEN_PORT=2222
|
||||
# Enable Gitea Actions (act_runner required on host)
|
||||
- GITEA__actions__ENABLED=true
|
||||
# Don't fetch actions from GitHub (offline mode + local only)
|
||||
- GITEA__actions__DEFAULT_ACTIONS_URL=off
|
||||
volumes:
|
||||
- /mnt/HoardingCow_docker_data/Gitea:/data
|
||||
networks:
|
||||
@@ -40,6 +46,22 @@ services:
|
||||
# Internal Routing
|
||||
- "traefik.http.services.gitea.loadbalancer.server.port=3000"
|
||||
|
||||
act_runner:
|
||||
image: gitea/act_runner:latest
|
||||
container_name: act_runner
|
||||
environment:
|
||||
- GITEA_INSTANCE_URL=https://code.lazyworkhorse.net
|
||||
- GITEA_RUNNER_REGISTRATION_TOKEN=${GITEA_RUNNER_TOKEN}
|
||||
- GITEA_RUNNER_NAME=ai-host-runner
|
||||
- GITEA_RUNNER_LABELS=ubuntu-latest:docker://catthehacker/ubuntu:full-22.04,nixos-builder:docker://nixos/nix
|
||||
volumes:
|
||||
- /var/run/docker.sock:/var/run/docker.sock
|
||||
networks:
|
||||
- vc_net
|
||||
restart: always
|
||||
depends_on:
|
||||
- gitea
|
||||
|
||||
networks:
|
||||
vc_net:
|
||||
external: true
|
||||
|
||||
Reference in New Issue
Block a user