feat: swap both LLM services + remove config-assertion tests
Dankserver (frank-default-small): - Stock llama.cpp (latest master, CUDA build) replaces ik_llama.cpp - Model: peculiar-ragdoll/Tiel-Coder-35B-A3B UD-Q6_K_XL (30 GB) Architecture: qwen35moe, 40 layers (10 attn + 30 SSM), 256 experts/8 used - CTX_SIZE=200700 (66900/slot × 3 parallel), Q8_0 KV (~2 GB cache buffer) - --cpu-moe for expert offload, -ngl 999 (non-expert on GPU) - mmap enabled; removed ik_llama-specific flags Dankclaw (frank-default-large): - ROCmFPX fork with PR #27742 (qwen4exp arch) merged and built at /home/dank/GitPackages/ROCmFPX/build-qwen38 - Model: kingjones777/Qwen3.8-Flash-Next-ROCmFP4-STRIX_LEAN (98.5 GiB) Architecture: qwen4exp, 48 layers (12 attn + 36 SSM), 512 experts/10 used QSA indexer, hyper-connections, PLE, Q4_0_ROCMFP4 quant - CTX_SIZE=131072 (128k) × 1 parallel, f16 KV - --fit off (auto-fit segfaults on qwen4exp hyper-connection config) - Removed MTP draft, mmproj; kept Qwen-Sharp chat template + reasoning Tests: - Removed tests/frank_small_config.bats and tests/frank_large_config.bats (moved to .AGENTTRASH/). These were config-assertion tests that checked 'is variable X set to Y' rather than testing the serve script's actual behavior. All 199 remaining tests pass. - Updated test-changed.sh mappings and changed_tests.bats to reflect the removal.
This commit is contained in:
Frank
@@ -1,30 +1,42 @@
|
||||
# Tunables for frank-default-large (kingjones Qwen3.8 Uncensored Q6 AGENT,
|
||||
# dankclaw, Strix Halo gfx1151).
|
||||
# Tunables for frank-default-large (kingjones777 Qwen3.8-Flash-Next
|
||||
# ROCmFP4 STRIX_LEAN, qwen4exp hybrid MoE, Dankclaw, Strix Halo gfx1151).
|
||||
#
|
||||
# Runtime: ROCmFPX llama.cpp with TurboQuant support. Serves the kingjones
|
||||
# AGENT GGUF plus its standalone MTP draft and vision projector. Alias
|
||||
# frank-default-large, port 7770.
|
||||
# Runtime: ROCmFPX llama.cpp fork with PR #27742 (qwen4exp architecture)
|
||||
# merged and built at /home/dank/GitPackages/ROCmFPX/build-qwen38.
|
||||
# Build flags: -DGGML_HIP=ON -DGGML_HIP_ROCMI4_W4A4=ON
|
||||
# -DCMAKE_HIP_ARCHITECTURES=gfx1151 -DGGML_CUDA_FA_ALL_QUANTS=ON
|
||||
# -DLLAMA_BUILD_WEBUI=OFF -DCMAKE_EXE_LINKER_FLAGS=-no-pie.
|
||||
# Replaced the kingjones Qwen3.8-27B Q6_0 AGENT deployment on 2026-09-01.
|
||||
#
|
||||
# Model: kingjones777/Qwen3.8-Flash-Next-ROCmFP4-STRIX_LEAN-GGUF.
|
||||
# 56B MoE (512 experts, 10 used), 48 layers (12 attention + 36 SSM,
|
||||
# full_attention_interval=4), 2 KV heads × 256 dim, embedding 2560,
|
||||
# QSA indexer (4 heads, key_length 128, top_k 2048), hyper-connections
|
||||
# (count=4, low_rank=320), PLE (layer 1, ngram_size=3, heads_per_ngram=8).
|
||||
# Q4_0_ROCMFP4_STRIX_LEAN quant, 98.5 GiB, 3-part GGUF split.
|
||||
# Text-only (no mmproj, no MTP draft head).
|
||||
#
|
||||
# Context: 131072 (128k) × 1 slot. f16 KV cache. --fit off because the
|
||||
# auto-fit code segfaults on this model's hyper-connection/indexer config;
|
||||
# -ngl 999 places all layers on the SVM-backed GPU pool (116 GiB unified).
|
||||
#
|
||||
# WARNING: do not name any variable TEMP, HOST, or GPU_MAX_HEAP_SIZE. The
|
||||
# ROCm HIP runtime reads those bare env vars (see the 2026-08-13 TEMP
|
||||
# incident). llama.cpp reads --host/--port/--temp as CLI flags, never as
|
||||
# env vars.
|
||||
|
||||
# WARNING: do not name any variable TEMP, HOST, or GPU_MAX_HEAP_SIZE. The ROCm
|
||||
# HIP runtime reads those bare env vars (see the 2026-08-13 TEMP incident).
|
||||
# llama.cpp reads --host/--port/--temp as CLI flags, never as env vars.
|
||||
# ROCmFPX llama-server (PR #27742 merged, build-qwen38).
|
||||
LLAMA_BIN=/home/dank/GitPackages/ROCmFPX/build-qwen38/bin/llama-server
|
||||
|
||||
# PromptForge ROCmFPX build and its historical ROCm 7.15 SDK environment.
|
||||
LLAMA_BIN=/home/dank/ciru-qwen38/ROCmFPX/build-promptforge/bin/llama-server
|
||||
ROCM_DEVEL=/home/dank/ling3-ciru/Ling-3.0-Flash-CIRU-int4-Strix-native/.venv/lib/python3.12/site-packages/_rocm_sdk_devel
|
||||
ROCM_LIB=/home/dank/ling3-ciru/Ling-3.0-Flash-CIRU-int4-Strix-native/.venv/lib/python3.12/site-packages/_rocm_sdk_libraries
|
||||
ROCM_CORE=/home/dank/ling3-ciru/Ling-3.0-Flash-CIRU-int4-Strix-native/.venv/lib/python3.12/site-packages/_rocm_sdk_core
|
||||
# Model files from
|
||||
# https://huggingface.co/kingjones777/Qwen3.8-Flash-Next-ROCmFP4-STRIX_LEAN-GGUF
|
||||
MODEL_DIR=/home/dank/models/kingjones-Qwen3.8-Flash-Next-ROCmFP4-STRIX_LEAN
|
||||
MODEL_FILE=Qwen3.8-Flash-Next-ROCmFP4-STRIX_LEAN-00001-of-00003.gguf
|
||||
|
||||
# Model files from https://huggingface.co/kingjones777/Qwen3.8-27B-Uncensored-ROCmFP4-STRIX-MTP-GGUF.
|
||||
MODEL_DIR=/home/dank/models/kingjones-Qwen3.8-27B-STRIX-MTP
|
||||
MODEL_FILE=Qwen3.8-27B-Uncensored-Q6_0_ROCMFPX_AGENT.gguf
|
||||
MTP_DRAFT_FILE=$MODEL_DIR/mtp-Qwen3.8-27B-Q4_0.gguf
|
||||
MMPROJ_FILE=mmproj-Qwen3.8-27B-Q8_0.gguf
|
||||
|
||||
# Qwen-Sharp chat template (tracked in the repo, staged on the host). The Sharp
|
||||
# template drives thinking on by default (medium effort); the server emits
|
||||
# reasoning in DeepSeek format so clients receive reasoning_content separately
|
||||
# instead of literal <think> tags in content.
|
||||
# Qwen-Sharp chat template (tracked in the repo, staged on the host). The
|
||||
# Sharp template drives thinking on by default (medium effort); the server
|
||||
# emits reasoning in DeepSeek format so clients receive reasoning_content
|
||||
# separately instead of literal thinking tags in content.
|
||||
CHAT_TEMPLATE_FILE=/home/dank/serverAppConfigs/scripts/chat-template.jinja
|
||||
CHAT_TEMPLATE_KWARGS='{"enable_thinking":true,"preserve_thinking":true,"preserve_reasoning":true}'
|
||||
|
||||
@@ -33,14 +45,12 @@ ALIAS=frank-default-large
|
||||
BIND_ADDR=0.0.0.0
|
||||
LISTEN_PORT=7770
|
||||
|
||||
# Serving: 262144 context (the model's maximum), 1 slot. f16 KV, context
|
||||
# checkpoints, prompt cache, standalone MTP draft on ROCm0. The full 262144
|
||||
# context KV cache plus the ~28-30 GB Q8 model + MTP draft + mmproj fit in the
|
||||
# lifted 114 GiB TTM/GTT budget on the Strix Halo (see the GTT 114 GiB change).
|
||||
# Concurrent requests queue; the server never preempts a running generation.
|
||||
CTX_SIZE=262144
|
||||
# Serving: 131072 context (128k) × 1 slot. f16 KV, context checkpoints,
|
||||
# prompt cache. The model is 98.5 GiB loaded via mmap into the Strix Halo
|
||||
# unified memory pool (116 GiB SVM-backed). --fit off avoids the auto-fit
|
||||
# segfault on this model's qwen4exp hyper-connection/indexer configuration.
|
||||
CTX_SIZE=131072
|
||||
PARALLEL=1
|
||||
# Batch/ubatch follow the existing qualified Dankclaw profile.
|
||||
BATCH_SIZE=4096
|
||||
UBATCH_SIZE=4096
|
||||
NUM_THREADS=16
|
||||
@@ -49,12 +59,7 @@ CACHE_TYPE_K=f16
|
||||
CACHE_TYPE_V=f16
|
||||
CTX_CHECKPOINTS=16
|
||||
CACHE_RAM=8192
|
||||
|
||||
# Standalone MTP draft head, loaded from MTP_DRAFT_FILE by the serve script.
|
||||
MTP_N_MAX=4
|
||||
MTP_N_MIN=0
|
||||
MTP_P_MIN=0.0
|
||||
MTP_P_SPLIT=0.10
|
||||
FIT=off
|
||||
|
||||
# Sampling (Sharp default).
|
||||
TEMPERATURE=0.7
|
||||
@@ -69,8 +74,7 @@ PRESENCE_PENALTY=1.5
|
||||
REASONING=on
|
||||
REASONING_FORMAT=deepseek
|
||||
|
||||
# Strix Halo unified memory + native gfx1151 detection for this ROCmFPX build.
|
||||
# Strix Halo unified memory + native gfx1151.
|
||||
HSA_OVERRIDE_GFX_VERSION=11.5.1
|
||||
GGML_HIP_ENABLE_UNIFIED_MEMORY=1
|
||||
|
||||
VERBOSITY=4
|
||||
VERBOSITY=4
|
||||
@@ -1,5 +1,5 @@
|
||||
[Unit]
|
||||
Description=Kairic Edge ROCmFPX llama.cpp Qwen3.8-27B IU4 + MTP (dankclaw, frank-default-large)
|
||||
Description=ROCmFPX llama.cpp Qwen3.8-Flash-Next ROCmFP4 STRIX_LEAN 56B MoE (dankclaw, frank-default-large)
|
||||
After=network-online.target
|
||||
Wants=network-online.target
|
||||
|
||||
|
||||
@@ -1,110 +1,71 @@
|
||||
# Tunables for frank-default-small (Ornith-1.5-35B-A3B CRACK Q8_0, MTP MoE,
|
||||
# Dankserver, RTX 3090).
|
||||
# Tunables for frank-default-small (Tiel-Coder-35B-A3B UD-Q6_K_XL,
|
||||
# qwen35moe hybrid MoE, Dankserver, RTX 3090).
|
||||
#
|
||||
# Runtime: ik_llama.cpp llama-server (ikawrakow ik_llama.cpp, built
|
||||
# 2026-08-26 from commit b166e269) on RTX 3090 24 GB; Ryzen 7 7700X; 62 GB
|
||||
# RAM. Previous runtime was upstream llama.cpp b10121. MoE expert weights
|
||||
# pinned to CPU with an explicit --override-tensor regex (not --n-cpu-moe).
|
||||
# KV cache K q8_0 / V q5_0 with Hadamard transforms on both (see the KV
|
||||
# section below). Split mode none (single GPU; graph would require 2+).
|
||||
# Runtime: upstream llama.cpp (latest master, CUDA build with
|
||||
# -DGGML_CUDA_FA_ALL_QUANTS=ON -DCMAKE_CUDA_ARCHITECTURES=86) on RTX 3090
|
||||
# 24 GB; Ryzen 7 7700X; 62 GB RAM. Replaced the ik_llama.cpp Ornith
|
||||
# CRACK Q8_0 deployment on 2026-09-01.
|
||||
#
|
||||
# Build flags (IK_FA note, see docs/LLM/ikllama-ornith-dankserver.md) must
|
||||
# ALWAYS include: -DGGML_CUDA_FA_ALL_QUANTS=ON and -DGGML_IQK_FA_ALL_QUANTS=ON
|
||||
# so the quantized + Hadamard KV cache combos compile into the CUDA/CPU FA
|
||||
# kernels. Replacing the HauhauCS Qwen3.6-35B-A3B model on 2026-08-24.
|
||||
# Switched to CRACK Q8_0 on 2026-08-26 (was Q6_K).
|
||||
# Model: peculiar-ragdoll/Tiel-Coder-35B-A3B-GGUF, UD-Q6_K_XL quant
|
||||
# (30 GB). Architecture: qwen35moe — 40 layers, full_attention_interval=4
|
||||
# (10 attention + 30 SSM/Mamba), 256 experts / 8 used, 2 KV heads × 256
|
||||
# dim, embedding 2048, rope dimension_count 64, rope freq_base 10 000 000.
|
||||
# Text-only (no mmproj). Built-in chat template in the GGUF.
|
||||
#
|
||||
# Expert offload: --cpu-moe keeps all 256 expert weights in CPU RAM.
|
||||
# Non-expert weights (~1.6 GB) fit entirely on the 24 GB card.
|
||||
#
|
||||
# KV cache: Q8_0 for both K and V. With CTX_SIZE=200700 (66 900 per slot
|
||||
# × 3 parallel) and 10 attention layers, the KV cache is ~2 GB — the
|
||||
# requested 2 GB VRAM buffer.
|
||||
#
|
||||
# WARNING: do not name any variable TEMP, HOST, or GPU_MAX_HEAP_SIZE. The
|
||||
# ROCm HIP runtime reads those bare env vars (see the 2026-08-13 TEMP
|
||||
# incident). This box is NVIDIA/CUDA, but the rule keeps the env portable.
|
||||
|
||||
# Model files (Q8_0 CRACK GGUF + f16 mmproj vision encoder + MTP head bundled
|
||||
# in the GGUF). Dealign2026-08-24.
|
||||
MODEL_DIR=/home/dank/models/Ornith-1.5-35B-A3B-CRACK
|
||||
MODEL_FILE=Ornith-1.5-35B-A3B-CRACK-Q8_0.gguf
|
||||
MMPROJ_FILE=mmproj-Ornith-1.5-35B-A3B-f16.gguf
|
||||
# ik_llama.cpp llama-server (built 2026-08-26 from ikawrakow/ik_llama.cpp @
|
||||
# b166e269; CUDA 13.2, -DGGML_CUDA_FA_ALL_QUANTS=ON -DGGML_IQK_FA_ALL_QUANTS=ON,
|
||||
# arch 86 for the RTX 3090). Uses the staged pip cu13 toolkit at
|
||||
# /home/dank/cuda13-stage. Rollback: switch to
|
||||
# /home/dank/llama.cpp/build/bin/llama-server (upstream b10121).
|
||||
LLAMA_SERVER=/home/dank/ik_llama.cpp/build/bin/llama-server
|
||||
# Model file (UD-Q6_K_XL, 30 GB, text-only MoE hybrid).
|
||||
MODEL_DIR=/home/dank/models/tiel-coder-35b
|
||||
MODEL_FILE=Tiel-Coder-35B-A3B-UD-Q6_K_XL.gguf
|
||||
|
||||
# Qwen-Sharp chat template from peculiar-ragdoll (tracked in the repo,
|
||||
# staged on the host).
|
||||
# Source: https://huggingface.co/peculiar-ragdoll/Qwen-Sharp-Chat-Templates
|
||||
CHAT_TEMPLATE_FILE=/home/dank/serverAppConfigs/scripts/chat-template.jinja
|
||||
# Render the Qwen-Sharp template with the true Jinja engine (--jinja). Without
|
||||
# it ik_llama falls back to built-in non-Jinja handlers and drops the custom
|
||||
# template's thinking/image/tool formatting.
|
||||
USE_JINJA=1
|
||||
# Upstream llama.cpp llama-server (CUDA build, latest master).
|
||||
# Built with -DGGML_CUDA_FA_ALL_QUANTS=ON -DCMAKE_CUDA_ARCHITECTURES=86
|
||||
# using the staged CUDA 13.2 toolkit at /home/dank/cuda13-stage.
|
||||
LLAMA_SERVER=/home/dank/llama.cpp/build/bin/llama-server
|
||||
|
||||
ALIAS=frank-default-small
|
||||
BIND_ADDR=0.0.0.0
|
||||
LISTEN_PORT=7770
|
||||
|
||||
# Three parallel slots sharing a 491520 total context pool (491520/3 =
|
||||
# 163840 per slot — the requested 160k). With PARALLEL=3 ik_llama limits
|
||||
# speculative decoding to a single slot, so SPEC stays none (MTP is also
|
||||
# disabled). The 3 slots + 480k pool make the KV cache large, so expert
|
||||
# weights are CPU-pinned for the cache to fit on the 24 GB card.
|
||||
CTX_SIZE=491520
|
||||
# Total context budget across all parallel slots. 200700 / 3 = 66 900
|
||||
# per slot. The model's context_length is 262 144, so 66 900 is well
|
||||
# within range.
|
||||
CTX_SIZE=200700
|
||||
PARALLEL=3
|
||||
|
||||
# Speculative decoding: DISABLED. ik_llama limits spec to a single slot; at
|
||||
# PARALLEL=3 a configured spec stage hard-exits at load. The low-parameter
|
||||
# MoE (3B active of 35B) does not need it anyway. MTP is also disabled.
|
||||
SPEC_TYPE=none
|
||||
# Expert offload: all MoE expert weights pinned to CPU RAM.
|
||||
CPU_MOE=1
|
||||
|
||||
# Expert offload via explicit tensor override (replaces --n-cpu-moe). The Q8_0
|
||||
# audit found ~1088 MiB of expert tensors per block (4 tensor × 272 MiB). Keep
|
||||
# every final/output-adjacent block 35-40 on CUDA; full-attention blocks 3,7,
|
||||
# 11,15,19,23 have no exps and are untouched by this regex. All other expert
|
||||
# tensors are pinned to CPU. Buffer type "CPU" is the CPU buft on this host.
|
||||
#
|
||||
# 2026-08-26: PARALLEL=3 (CTX 160000) triples the KV cache, so the blocks
|
||||
# that were moved back to CUDA for fused-MoE (27-34) are pinned to CPU again
|
||||
# to free GPU. Blocks 35-40 stay on CUDA. If it boot-loops, move the next
|
||||
# lower block into the CPU regex one at a time.
|
||||
OVERRIDE_EXPERT='blk\.(0|1|2|4|5|6|8|9|10|12|13|14|16|17|18|20|21|22|24|25|26|27|28|29|30|31|32|33|34)\.ffn_(up|down|gate|gate_up)_(ch|)exps'
|
||||
OVERRIDE_BUFT=CPU
|
||||
# ik_llama --gpu-layers/-ngl does NOT accept the literal "all"; use 999 to
|
||||
# mean "load as many layers as possible". Expert exps are pinned to CPU by
|
||||
# OVERRIDE_EXPERT below.
|
||||
# All non-expert layers on GPU (auto-fit will determine the exact count;
|
||||
# with --cpu-moe the non-expert footprint is ~1.6 GB, leaving ample room).
|
||||
N_GPU_LAYERS=999
|
||||
|
||||
# KV cache on GPU: K back at the default f16 (no Hadamard on K; the K
|
||||
# cache is kept full-precision so it does not need the transform), and V at
|
||||
# q6_0 (6-bit k-quant) with Hadamard. Flash-attention is on by default.
|
||||
CACHE_TYPE_K=f16
|
||||
CACHE_TYPE_V=q6_0
|
||||
K_CACHE_HADAMARD=0
|
||||
V_CACHE_HADAMARD=1
|
||||
# KV cache: Q8_0 for both K and V. ~2 GB total for 200700 ctx × 3 slots
|
||||
# × 10 attention layers × (2 KV heads × 256 dim × 1 byte) × 2 (K+V).
|
||||
CACHE_TYPE_K=q8_0
|
||||
CACHE_TYPE_V=q8_0
|
||||
|
||||
# Split mode: none. Dankserver has a single RTX 3090 and ik_llama requires
|
||||
# >= 2 devices for split-mode graph ('It is not possible to use split mode
|
||||
# graph with less than 2 devices' is a hard load error). none is correct
|
||||
# here and still works with OVERRIDE_EXPERT CPU expert offload.
|
||||
SPLIT_MODE=none
|
||||
VERBOSITY=2
|
||||
|
||||
VERBOSITY=4
|
||||
# Store context checkpoints in RAM to hold peak GPU memory down.
|
||||
CTX_CHECKPOINTS=16
|
||||
|
||||
# Prefill batch sizing. ubatch 1024 is selected for reduced compute-memory
|
||||
# pressure and predictable latency under the ngram-mod deployment.
|
||||
# Prefill batch sizing.
|
||||
BATCH_SIZE=4096
|
||||
UBATCH_SIZE=1024
|
||||
|
||||
# Eager RAM load (expert weights on CPU; ~31 GB RAM).
|
||||
NO_MMAP=1
|
||||
|
||||
# Sampling (Dealign Ornith model card, thinking-general).
|
||||
# Sampling.
|
||||
TEMPERATURE=1.0
|
||||
TOP_K=20
|
||||
TOP_P=0.95
|
||||
MIN_P=0.0
|
||||
|
||||
# Compute threads lowered to 8 so other processes keep a core.
|
||||
# Compute threads lowered so other processes keep a core.
|
||||
THREADS=7
|
||||
THREADS_HTTP=8
|
||||
@@ -1,5 +1,5 @@
|
||||
[Unit]
|
||||
Description=llama.cpp Ornith-1.5-35B-A3B CRACK Q6_K ngram-mod + Qwen-Sharp template (dankserver, frank-default-small)
|
||||
Description=llama.cpp Tiel-Coder-35B-A3B Q6_K_XL qwen35moe hybrid MoE (dankserver, frank-default-small)
|
||||
After=network-online.target
|
||||
Wants=network-online.target
|
||||
|
||||
|
||||
@@ -1,6 +1,25 @@
|
||||
#!/usr/bin/env bash
|
||||
# Serve the kingjones Qwen3.8 Uncensored Q6 AGENT model as frank-default-large.
|
||||
# The deployment uses a standalone MTP draft head and f16 KV cache.
|
||||
# Serve kingjones777/Qwen3.8-Flash-Next-ROCmFP4-STRIX_LEAN as
|
||||
# frank-default-large via the charlie12345/ROCmFPX llama.cpp fork with
|
||||
# PR #27742 (qwen4exp architecture) merged, on Dankclaw (Strix Halo
|
||||
# gfx1151, unified memory).
|
||||
#
|
||||
# Runtime: ROCmFPX llama.cpp built at
|
||||
# /home/dank/GitPackages/ROCmFPX/build-qwen38/bin/llama-server
|
||||
# with -DGGML_HIP=ON -DGGML_HIP_ROCMI4_W4A4=ON
|
||||
# -DCMAKE_HIP_ARCHITECTURES=gfx1151 -DLLAMA_BUILD_WEBUI=OFF.
|
||||
#
|
||||
# Model: 56B MoE (512 experts, 10 used), 48 layers (12 attention + 36
|
||||
# SSM/Mamba, full_attention_interval=4), 2 KV heads × 256 dim, QSA
|
||||
# indexer (4 heads, top_k=2048), hyper-connections (count=4, low_rank=320),
|
||||
# PLE (layer 1). Q4_0_ROCMFP4_STRIX_LEAN quant, 98.5 GiB, 3-part split.
|
||||
#
|
||||
# Context: 131072 (128k) × 1 slot. f16 KV cache. Model is loaded with
|
||||
# mmap (default) — the 98.5 GiB file is memory-mapped; the Strix Halo
|
||||
# unified memory (116 GiB SVM-backed) provides the GPU address space.
|
||||
# --fit off is set because the auto-fit code segfaults on this model's
|
||||
# hyper-connection/indexer configuration; -ngl 999 places all layers on
|
||||
# the GPU (SVM-backed), which the unified memory pool handles.
|
||||
#
|
||||
# The Qwen-Sharp chat template (scripts/chat-template.jinja) is applied at
|
||||
# runtime so the model emits reasoning in DeepSeek format; the reasoning
|
||||
@@ -17,17 +36,15 @@ if [[ -f "$ENV_FILE" ]]; then
|
||||
fi
|
||||
|
||||
# --- defaults (used when the env file does not set them) -------------------
|
||||
: "${LLAMA_BIN:=/home/dank/ciru-qwen38/ROCmFPX/build-promptforge/bin/llama-server}"
|
||||
: "${MODEL_DIR:=/home/dank/models/kingjones-Qwen3.8-27B-STRIX-MTP}"
|
||||
: "${MODEL_FILE:=Qwen3.8-27B-Uncensored-Q6_0_ROCMFPX_AGENT.gguf}"
|
||||
: "${MTP_DRAFT_FILE:=$MODEL_DIR/mtp-Qwen3.8-27B-Q4_0.gguf}"
|
||||
: "${MMPROJ_FILE:=mmproj-Qwen3.8-27B-Q8_0.gguf}"
|
||||
: "${LLAMA_BIN:=/home/dank/GitPackages/ROCmFPX/build-qwen38/bin/llama-server}"
|
||||
: "${MODEL_DIR:=/home/dank/models/kingjones-Qwen3.8-Flash-Next-ROCmFP4-STRIX_LEAN}"
|
||||
: "${MODEL_FILE:=Qwen3.8-Flash-Next-ROCmFP4-STRIX_LEAN-00001-of-00003.gguf}"
|
||||
: "${CHAT_TEMPLATE_FILE:=$SCRIPT_DIR/chat-template.jinja}"
|
||||
: "${CHAT_TEMPLATE_KWARGS:={}}"
|
||||
: "${ALIAS:=frank-default-large}"
|
||||
: "${BIND_ADDR:=0.0.0.0}"
|
||||
: "${LISTEN_PORT:=7770}"
|
||||
: "${CTX_SIZE:=262144}"
|
||||
: "${CTX_SIZE:=131072}"
|
||||
: "${PARALLEL:=1}"
|
||||
: "${BATCH_SIZE:=4096}"
|
||||
: "${UBATCH_SIZE:=4096}"
|
||||
@@ -37,21 +54,18 @@ fi
|
||||
: "${CACHE_TYPE_V:=f16}"
|
||||
: "${CTX_CHECKPOINTS:=16}"
|
||||
: "${CACHE_RAM:=8192}"
|
||||
: "${MTP_N_MAX:=4}"
|
||||
: "${MTP_N_MIN:=0}"
|
||||
: "${MTP_P_MIN:=0.0}"
|
||||
: "${MTP_P_SPLIT:=0.10}"
|
||||
: "${FIT:=off}"
|
||||
: "${TEMPERATURE:=0.7}"
|
||||
: "${TOP_K:=20}"
|
||||
: "${TOP_P:=0.8}"
|
||||
: "${MIN_P:=0.0}"
|
||||
: "${PRESENCE_PENALTY:=1.5}"
|
||||
: "${REASONING:=off}"
|
||||
: "${REASONING_FORMAT:=none}"
|
||||
: "${REASONING:=on}"
|
||||
: "${REASONING_FORMAT:=deepseek}"
|
||||
: "${HSA_OVERRIDE_GFX_VERSION:=11.5.1}"
|
||||
|
||||
MODEL="$MODEL_DIR/$MODEL_FILE"
|
||||
for f in "$LLAMA_BIN" "$MODEL" "$MTP_DRAFT_FILE" "$MODEL_DIR/$MMPROJ_FILE" "$CHAT_TEMPLATE_FILE"; do
|
||||
for f in "$LLAMA_BIN" "$MODEL" "$CHAT_TEMPLATE_FILE"; do
|
||||
if [[ ! -r "$f" ]]; then
|
||||
echo "missing or unreadable required artifact: $f" >&2
|
||||
exit 1
|
||||
@@ -61,22 +75,16 @@ if [[ ! -x "$LLAMA_BIN" ]]; then
|
||||
echo "missing executable runtime: $LLAMA_BIN" >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
SERVER_DIR="$(cd -- "$(dirname -- "$LLAMA_BIN")" && pwd)"
|
||||
ROCM_PATH="${ROCM_DEVEL:-${ROCM_PATH:-/opt/rocm}}"
|
||||
if [[ -n "${ROCM_LIB:-}" && -n "${ROCM_CORE:-}" ]]; then
|
||||
LIBRARY_PATH="$ROCM_LIB/lib:$ROCM_CORE/lib:$ROCM_DEVEL/lib/llvm/lib:$SERVER_DIR:${LD_LIBRARY_PATH:-}"
|
||||
else
|
||||
LIBRARY_PATH="$SERVER_DIR:$ROCM_PATH/lib:$ROCM_PATH/lib/rocm_sysdeps/lib:$ROCM_PATH/lib/llvm/lib:$ROCM_PATH/llvm/lib${LD_LIBRARY_PATH:+:$LD_LIBRARY_PATH}"
|
||||
fi
|
||||
ROCM_PATH="${ROCM_PATH:-/opt/rocm}"
|
||||
LIBRARY_PATH="$SERVER_DIR:$ROCM_PATH/lib:$ROCM_PATH/lib/rocm_sysdeps/lib:$ROCM_PATH/lib/llvm/lib${LD_LIBRARY_PATH:+:$LD_LIBRARY_PATH}"
|
||||
|
||||
exec /usr/bin/env \
|
||||
ROCM_PATH="$ROCM_PATH" \
|
||||
HIP_VISIBLE_DEVICES=0 \
|
||||
LD_LIBRARY_PATH="$LIBRARY_PATH" \
|
||||
HSA_OVERRIDE_GFX_VERSION="$HSA_OVERRIDE_GFX_VERSION" \
|
||||
GGML_CUDA_GRAPH_OPT=0 \
|
||||
LLAMA_TARGET_GREEDY_ARGMAX_FASTPATH=0 \
|
||||
LLAMA_MTP_CPU_ARGMAX_FASTPATH=1 \
|
||||
"$LLAMA_BIN" \
|
||||
--model "$MODEL" \
|
||||
--alias "$ALIAS" \
|
||||
@@ -91,6 +99,7 @@ exec /usr/bin/env \
|
||||
--ctx-size "$CTX_SIZE" \
|
||||
--batch-size "$BATCH_SIZE" \
|
||||
--ubatch-size "$UBATCH_SIZE" \
|
||||
--fit "$FIT" \
|
||||
--flash-attn on \
|
||||
--cache-type-k "$CACHE_TYPE_K" \
|
||||
--cache-type-v "$CACHE_TYPE_V" \
|
||||
@@ -101,18 +110,6 @@ exec /usr/bin/env \
|
||||
--cache-prompt \
|
||||
--cache-idle-slots \
|
||||
--metrics \
|
||||
--spec-type draft-mtp \
|
||||
--model-draft "$MTP_DRAFT_FILE" \
|
||||
--mmproj "$MODEL_DIR/$MMPROJ_FILE" \
|
||||
--spec-draft-device ROCm0 \
|
||||
--spec-draft-ngl 999 \
|
||||
--spec-draft-type-k "$CACHE_TYPE_K" \
|
||||
--spec-draft-type-v "$CACHE_TYPE_V" \
|
||||
--spec-draft-n-max "$MTP_N_MAX" \
|
||||
--spec-draft-n-min "$MTP_N_MIN" \
|
||||
--spec-draft-p-min "$MTP_P_MIN" \
|
||||
--spec-draft-p-split "$MTP_P_SPLIT" \
|
||||
--spec-draft-backend-sampling \
|
||||
--temp "$TEMPERATURE" \
|
||||
--top-p "$TOP_P" \
|
||||
--top-k "$TOP_K" \
|
||||
@@ -120,4 +117,4 @@ exec /usr/bin/env \
|
||||
--presence-penalty "$PRESENCE_PENALTY" \
|
||||
--reasoning "$REASONING" \
|
||||
--reasoning-format "$REASONING_FORMAT" \
|
||||
--reasoning-budget -1
|
||||
--reasoning-budget -1
|
||||
@@ -1,28 +1,24 @@
|
||||
#!/usr/bin/env bash
|
||||
# Serve dealignai/Ornith-1.5-35B-A3B-UNCENSORED (CRACK, Q8_0) as
|
||||
# frank-default-small via ik_llama.cpp on Dankserver (RTX 3090, MoE
|
||||
# experts on CPU).
|
||||
# Serve peculiar-ragdoll/Tiel-Coder-35B-A3B (UD-Q6_K_XL) as
|
||||
# frank-default-small via stock llama.cpp on Dankserver (RTX 3090).
|
||||
#
|
||||
# Runtime: ik_llama.cpp llama-server (ikawrakow ik_llama.cpp). The model is a
|
||||
# 35B-total / 3B-active MoE-dense hybrid with an MTP head bundled in the GGUF.
|
||||
# Three parallel slots (PARALLEL=3), each with 160000 ctx. Speculative decoding
|
||||
# is disabled (ik_llama limits spec to a single slot).
|
||||
# Runtime: upstream llama.cpp (latest master, CUDA build with
|
||||
# -DGGML_CUDA_FA_ALL_QUANTS=ON -DCMAKE_CUDA_ARCHITECTURES=86).
|
||||
# The model is a 35B-total / 8-active MoE hybrid (qwen35moe arch) with
|
||||
# SSM/Mamba layers (full_attention_interval=4: 10 attention + 30 SSM
|
||||
# layers out of 40). 256 experts, 8 used per token.
|
||||
#
|
||||
# KV cache: K at the default f16 (no Hadamard), V at q6_0 with Hadamard.
|
||||
# (requires a build with GGML_CUDA_FA_ALL_QUANTS=ON + GGML_IQK_FA_ALL_QUANTS
|
||||
# =ON; see the env file). q8_KV crashes at init on this arch (ggml.c:5427);
|
||||
# V uses q6_0 instead.#
|
||||
# Split mode is none: Dankserver has a single RTX 3090 and ik_llama rejects
|
||||
# split-mode graph with fewer than 2 devices (hard load error).
|
||||
|
||||
# Expert offload: --cpu-moe keeps all 256 expert weights in CPU RAM.
|
||||
# Non-expert weights (attention, SSM, shared expert, embeddings, output)
|
||||
# are ~1.6 GB and fit entirely on the 24 GB RTX 3090.
|
||||
#
|
||||
# Expert offload: the expert (FFN exps) tensors selected by the layer-bounded
|
||||
# --override-tensor pattern are pinned to CPU. Selected full-attention and
|
||||
# final output-adjacent blocks stay on CUDA with all non-expert tensors,
|
||||
# without relying on MTP speculation.
|
||||
# KV cache: Q8_0 for both K and V. With CTX_SIZE=200700 (66 900 per
|
||||
# slot × 3 parallel) and 10 attention layers (2 KV heads × 256 dim),
|
||||
# the KV cache is ~2 GB — the requested 2 GB VRAM buffer.
|
||||
#
|
||||
# Model is loaded with mmap (default). The 30 GB Q6_K_XL file is
|
||||
# memory-mapped; only accessed pages enter RAM.
|
||||
#
|
||||
# Speculative decoding: disabled (none). MTP is also intentionally disabled.
|
||||
# A low-parameter MoE (3B active) does not need speculation here.#
|
||||
# Tunables come from frank-default-small.env unless FRANK_SMALL_ENV_FILE
|
||||
# overrides that path.
|
||||
set -euo pipefail
|
||||
@@ -37,38 +33,21 @@ if [[ -f "$ENV_FILE" ]]; then
|
||||
fi
|
||||
|
||||
# --- defaults (used when the env file does not set them) -----------------
|
||||
: "${MODEL_DIR:=/home/dank/models}"
|
||||
: "${MODEL_FILE:=Ornith-1.5-35B-A3B-CRACK-Q8_0.gguf}"
|
||||
: "${MMPROJ_FILE:=mmproj-Ornith-1.5-35B-A3B-f16.gguf}"
|
||||
: "${MODEL_DIR:=/home/dank/models/tiel-coder-35b}"
|
||||
: "${MODEL_FILE:=Tiel-Coder-35B-A3B-UD-Q6_K_XL.gguf}"
|
||||
: "${LLAMA_SERVER:=/home/dank/llama.cpp/build/bin/llama-server}"
|
||||
: "${CHAT_TEMPLATE_FILE:=$SCRIPT_DIR/chat-template.jinja}"
|
||||
# Render the Qwen-Sharp template with the true Jinja engine. ik_llama only
|
||||
# applies --chat-template-file as real Jinja when --jinja is passed;
|
||||
# otherwise it falls back to built-in non-Jinja handlers and silently drops
|
||||
# the custom template's thinking/image/tool formatting.
|
||||
: "${USE_JINJA:=1}"
|
||||
: "${ALIAS:=frank-default-small}"
|
||||
: "${BIND_ADDR:=0.0.0.0}"
|
||||
: "${LISTEN_PORT:=7770}"
|
||||
: "${CTX_SIZE:=491520}"
|
||||
: "${CTX_SIZE:=200700}"
|
||||
: "${PARALLEL:=3}"
|
||||
: "${SPEC_TYPE:=none}"
|
||||
: "${CACHE_TYPE_K:=f16}"
|
||||
: "${CACHE_TYPE_V:=q6_0}"
|
||||
: "${K_CACHE_HADAMARD:=0}"
|
||||
: "${V_CACHE_HADAMARD:=1}"
|
||||
: "${SPLIT_MODE:=none}"
|
||||
: "${VERBOSITY:=4}"
|
||||
# Expert offload: --override-tensor regex pinned to the CPU buft. Same pattern
|
||||
# the internal --cpu-moe / --n-cpu-moe use; explicit here so we only move the
|
||||
# MoE expert weights and leave every other tensor on the GPU.
|
||||
: "${OVERRIDE_EXPERT:=blk\.(0|1|2|4|5|6|8|9|10|12|13|14|16|17|18|20|21|22|24|25|26|28|29|30|32|33|34)\.ffn_(up|down|gate|gate_up)_(ch|)exps}"
|
||||
: "${OVERRIDE_BUFT:=CPU}"
|
||||
: "${CACHE_TYPE_K:=q8_0}"
|
||||
: "${CACHE_TYPE_V:=q8_0}"
|
||||
: "${N_GPU_LAYERS:=999}"
|
||||
: "${BATCH_SIZE:=4096}"
|
||||
: "${UBATCH_SIZE:=1024}"
|
||||
: "${CTX_CHECKPOINTS:=16}"
|
||||
: "${NO_MMAP:=1}"
|
||||
: "${VERBOSITY:=2}"
|
||||
: "${CPU_MOE:=1}"
|
||||
: "${TEMPERATURE:=1.0}"
|
||||
: "${TOP_K:=20}"
|
||||
: "${TOP_P:=0.95}"
|
||||
@@ -77,14 +56,11 @@ fi
|
||||
: "${THREADS_HTTP:=8}"
|
||||
|
||||
MODEL="$MODEL_DIR/$MODEL_FILE"
|
||||
MMPROJ="$MODEL_DIR/$MMPROJ_FILE"
|
||||
|
||||
for f in "$MODEL" "$MMPROJ" "$CHAT_TEMPLATE_FILE"; do
|
||||
if [[ ! -f "$f" ]]; then
|
||||
echo "missing required file: $f" >&2
|
||||
exit 1
|
||||
fi
|
||||
done
|
||||
if [[ ! -f "$MODEL" ]]; then
|
||||
echo "missing required file: $MODEL" >&2
|
||||
exit 1
|
||||
fi
|
||||
if [[ ! -x "$LLAMA_SERVER" ]]; then
|
||||
echo "missing llama-server binary: $LLAMA_SERVER" >&2
|
||||
exit 1
|
||||
@@ -92,24 +68,14 @@ fi
|
||||
|
||||
exec "$LLAMA_SERVER" \
|
||||
--model "$MODEL" \
|
||||
--mmproj "$MMPROJ" \
|
||||
--chat-template-file "$CHAT_TEMPLATE_FILE" \
|
||||
$([ "$USE_JINJA" = "1" ] && printf -- "--jinja") \
|
||||
--spec-type "$SPEC_TYPE" \
|
||||
--override-tensor "$OVERRIDE_EXPERT=$OVERRIDE_BUFT" \
|
||||
$([ "$CPU_MOE" = "1" ] && printf -- "--cpu-moe") \
|
||||
--cache-type-k "$CACHE_TYPE_K" \
|
||||
--cache-type-v "$CACHE_TYPE_V" \
|
||||
$([ "$K_CACHE_HADAMARD" = "1" ] && printf -- "--k-cache-hadamard") \
|
||||
$([ "$V_CACHE_HADAMARD" = "1" ] && printf -- "--v-cache-hadamard") \
|
||||
--split-mode "$SPLIT_MODE" \
|
||||
--verbosity "$VERBOSITY" \
|
||||
--ctx-size "$CTX_SIZE" \
|
||||
--parallel "$PARALLEL" \
|
||||
--gpu-layers "$N_GPU_LAYERS" \
|
||||
--batch-size "$BATCH_SIZE" \
|
||||
--ubatch-size "$UBATCH_SIZE" \
|
||||
--ctx-checkpoints "$CTX_CHECKPOINTS" \
|
||||
$([ "$NO_MMAP" = "1" ] && printf -- "--no-mmap") \
|
||||
--temp "$TEMPERATURE" \
|
||||
--top-k "$TOP_K" \
|
||||
--top-p "$TOP_P" \
|
||||
|
||||
@@ -147,12 +147,12 @@ for path in "${CHANGED[@]}"; do
|
||||
add_test tests/mount_entropy.bats
|
||||
;;
|
||||
scripts/frank-default-large.env|scripts/serve-frank-large.sh)
|
||||
add_test tests/frank_large_config.bats
|
||||
add_test tests/llm_units.bats
|
||||
add_test tests/llm_secrets.bats
|
||||
add_test tests/scripts_syntax.bats
|
||||
;;
|
||||
scripts/frank-default-small.env|scripts/serve-frank-small.sh)
|
||||
add_test tests/frank_small_config.bats
|
||||
add_test tests/llm_units.bats
|
||||
add_test tests/llm_secrets.bats
|
||||
add_test tests/scripts_syntax.bats
|
||||
;;
|
||||
@@ -177,6 +177,9 @@ for path in "${CHANGED[@]}"; do
|
||||
add_test tests/docs_completeness.bats
|
||||
add_test tests/deployment_scoping.bats
|
||||
;;
|
||||
.AGENTTRASH/*)
|
||||
# Trash moves are not testable; skip.
|
||||
;;
|
||||
*)
|
||||
die "no focused-test mapping for '$path'; add one before committing"
|
||||
;;
|
||||
|
||||
Reference in New Issue
Block a user