iop/agent-test/dev/inventory.yaml
toki 367e7f3dc4 feat(openai): 모델 생성 정책을 적용한다
provider pool 모델 catalog의 출력 토큰 기본값과 thinking budget을 OpenAI-compatible 요청에 반영하고, Ornith dev-runtime 기준과 관련 계약 문서를 함께 맞춘다.
2026-07-04 09:20:56 +09:00

129 lines
4 KiB
YAML

test_env: dev
profile: dev-runtime-provider-pool
last_updated_at: "2026-07-03"
source:
remote_runner:
ssh: toki@toki-labs.com
repo_root: /Users/toki/agent-work/iop-dev
clean_sync:
- git fetch origin main
- git reset --hard origin/main
- git clean -fd
dirty_policy: discard
edge:
id: edge-toki-labs-dev
config_path: build/dev-runtime/edge.yaml
control_plane_http: http://127.0.0.1:18001
control_plane_status_url: http://127.0.0.1:18001/edges/edge-toki-labs-dev/status
bootstrap_http_public: http://toki-labs.com:18082
openai_base_url_public: http://toki-labs.com:18083/v1
openai_base_url_runner: http://127.0.0.1:18083/v1
edge_node_tcp_public: toki-labs.com:18084
admin_addr_runner: 127.0.0.1:19093
build:
binaries:
edge: build/dev-runtime/bin/edge
node_macos: build/dev-runtime/bin/iop-node
node_linux_arm64: build/dev-runtime/bin/iop-node-linux-arm64
node_windows_amd64: build/dev-runtime/bin/iop-node-windows-amd64.exe
model:
alias: ornith:35b
provider_capacity_total: 9
context_window: 262144
default_max_tokens: 32768
min_max_tokens: 32768
default_thinking_token_budget: 8192
capacity_smoke:
endpoints:
- /v1/responses
- /v1/chat/completions
concurrent_requests: capacity_plus_one
expected_total_in_flight: 9
expected_min_queued: 1
extended_concurrent_requests: 13
extended_expected_min_queued: 4
prompt_policy: long_reasoning_allowed
exact_output_match: false
nodes:
- id: mac-codex-node
alias: mac-codex
role: cli+mlx-provider
ssh: toki@toki-labs.com
workspace: /Users/toki/agent-work/iop-workspace/nomadcode
provider_pool_candidate: true
adapters:
- cli
- mac-mlx-vllm
providers:
- id: mac-mlx-vllm
type: vllm-mlx
endpoint: http://127.0.0.1:8002/v1
served_model: mlx-community/Ornith-1.0-35B-4bit
capacity: 2
runtime:
api_key_policy: local_bearer_from_edge_yaml
workdir: /Users/toki/agent-work/iop-mlx-vllm
python: .venv/bin/python
package_baseline:
vllm_mlx: 0.3.0
mlx: 0.31.2
mlx_lm: 0.31.3
model_cache: /Users/toki/agent-work/iop-mlx-vllm/hf-cache
pid_file: /Users/toki/agent-work/iop-mlx-vllm/vllm-mlx.pid
stdout_log: /Users/toki/agent-work/iop-mlx-vllm/logs/vllm-mlx.stdout.log
stderr_log: /Users/toki/agent-work/iop-mlx-vllm/logs/vllm-mlx.stderr.log
max_num_seqs: 2
max_kv_size: 262144
max_request_tokens: 262144
default_max_tokens: 32768
default_thinking_token_budget: 8192
paged_cache_block_size: 64
max_cache_blocks: 4096
total_kv_tokens: 262144
default_chat_template_kwargs:
enable_thinking: true
- id: gx10-vllm-node
alias: gx10-vllm
role: vllm-provider
ssh: toki@192.168.0.91
workspace: /home/toki/iop-gx10-vllm
provider_pool_candidate: true
provider:
id: gx10-vllm
type: vllm
endpoint: http://192.168.0.91:8001/v1
served_model: sakamakismile/Ornith-1.0-35B-NVFP4
capacity: 4
runtime:
max_model_len: 262144
max_num_seqs: 4
gpu_memory_utilization: 0.30
default_chat_template_kwargs:
enable_thinking: true
- id: onexplayer-lemonade-node
alias: onexplayer-lemonade
role: lemonade-provider
ssh: r0bin@192.168.0.59
ssh_origin: current_host
workspace: C:/Users/r0bin/iop-field
provider_pool_candidate: true
provider:
id: onexplayer-lemonade
type: lemonade
endpoint: http://192.168.0.59:13305/v1
served_model: Ornith-1.0-35B-GGUF-Q4_K_M
capacity: 3
load:
endpoint: http://192.168.0.59:13305/v1/load
backend: vulkan
ctx_size: 524288
llamacpp_args: "--spec-type none -np 3 -cb -fa on -b 4096 -ub 1024"
observed_slot_n_ctx: 174848
save_options: true
mtp_runtime_policy: disabled
windows_process_start: Win32_Process.Create