Instructions to use krush420/qwen3-coder-30b-sre-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use krush420/qwen3-coder-30b-sre-lora with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-Coder-30B-A3B-Instruct")
model = PeftModel.from_pretrained(base_model, "krush420/qwen3-coder-30b-sre-lora")

Transformers

How to use krush420/qwen3-coder-30b-sre-lora with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="krush420/qwen3-coder-30b-sre-lora")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("krush420/qwen3-coder-30b-sre-lora", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use krush420/qwen3-coder-30b-sre-lora with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "krush420/qwen3-coder-30b-sre-lora"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "krush420/qwen3-coder-30b-sre-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/krush420/qwen3-coder-30b-sre-lora

SGLang

How to use krush420/qwen3-coder-30b-sre-lora with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "krush420/qwen3-coder-30b-sre-lora" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "krush420/qwen3-coder-30b-sre-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "krush420/qwen3-coder-30b-sre-lora" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "krush420/qwen3-coder-30b-sre-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use krush420/qwen3-coder-30b-sre-lora with Docker Model Runner:
```
docker model run hf.co/krush420/qwen3-coder-30b-sre-lora
```

Qwen3-Coder-30B-A3B — KruschStack Wave-4 Unified Agentic SRE & Tool-Caller LoRA (`L2` Student)

Trained Through KruschStack to Benefit KruschStack

This Wave-4 Unified adapter is the official L2 Local MoE Specialist (3.3B active parameters) for KruschStack (krusch-stack) — the Unified 7-Pillar, 39-Tool Model Context Protocol (MCP) & Agentic Coding Stack (npx krusch-stack init). Combining AST-grounded SRE code auditing, @krusch/toolkit Lego-First synthesis, and multi-turn MCP/Workspace tool-calling with a role-neutral Modelfile, it beats Stock Qwen3-Coder-30B-A3B-Instruct on the full 600-task DolphinBench suite (42.67% [256/600] vs. 42.00% [252/600] under official gpt-5.6-sol grading, 85.73% [1,123/1,310] mechanical checks, $0.00 local inference cost) while achieving 100% Wave-4 Referee CLEAN_PASS + P0/P1 calibration.

🔗 The KruschStack & Jean SRE Ecosystem

🛠️ KruschStack (7-Pillar 39-Tool MCP Stack): github.com/kruschdev/krusch-stack (npx krusch-stack init)
🤖 Jean SRE by Kruschdev (24/7 Autonomous Auditor & Distillation Harness): github.com/kruschdev/Jean-SRE-by-kruschdev
📊 Training Dataset (18,131 Zero-Leak SFT + DPO Records across 7 Splits): krush420/sovereign-code-audit-sft
🏆 Hermetic Benchmark Audit Logs (68/68 Gates + DolphinBench 600): krush420/sovereign-g4-benchmarks
🪶 Companion 8B Edge Agent LoRA (L1.5 Tier): krush420/llama3-hermes3-8b-sre-lora (Built with Meta Llama 3 for NousResearch/Hermes-3-Llama-3.1-8B)
🧠 Companion 120B Teacher QLoRA (L2.5 Burst Tier): krush420/gpt-oss-120b-sre-lora (for openai/gpt-oss-120b)

🏗️ Built With & For the KruschStack 7-Pillar Architecture

qwen3-coder-30b-sre-lora reasons over structured tool outputs from the KruschStack Unified MCP Gateway and acts as the L2 Local MoE Specialist in krusch-cascade-router:

┌────────────────────────────────────────────────────────────────────────────┐
│                        KRUSCH-STACK UNIFIED GATEWAY                        │
│               `npx krusch-stack mcp`  (39 Annotated MCP Tools)             │
├──────────────────────┬───────────────────────────┬─────────────────────────┤
│ 1. WORKING MEMORY    │ 2A. CODEBASE AST & GIT    │ 2B. OS PAGE MEMORY      │
│ krusch-context-mcp   │ krusch-git                │ krusch-fs               │
│ • retrieve / remember│ • search_symbols          │ • locate / search       │
│ • revise / nudge     │ • dependency_graph        │ • page_tree / read_page │
│ • INV-12 span drift  │ • read_tree / read_blob   │ • remember_on_page      │
├──────────────────────┼───────────────────────────┼─────────────────────────┤
│ 3. 2PC HARNESS       │ 4. FRONTEND DESIGN MCP    │ 5. WORKSPACE HUB        │
│ krusch-harness       │ krusch-frontend-mcp       │ g-krusch                │
│ • run (stages diff)  │ • generate_color_palette  │ • get_agent_steering    │
│ • apply_diff (2PC)   │ • css_to_tailwind         │ • search_drive_rag      │
│ • explain / reject   │ • canonical_templates     │ • write (.gdoc/.slides) │
├──────────────────────┴───────────────────────────┴─────────────────────────┤
│ 6. COGNITIVE CASCADE ROUTER (`krusch-cascade-router`)                      │
│ • <25µs L1 fast-path -> [L2: qwen3-coder-30b-sre-lora] -> L2.5 120B        │
└────────────────────────────────────────────────────────────────────────────┘

What Wave-4 Unified SFT + DPO Adds Over Wave-3

Role-Neutral Modelfile + Multi-Turn <tool_call> Agency (agentic_tool_calling_sft & dpo): Removes Wave-3's hardcoded auditor SYSTEM prompt and trains directly on structured tools + assistant.tool_calls + role: "tool" chains covering read-before-write discovery (list_* -> update_*), temporal state supersession, and decisive side-effect execution.
Gemini 4 Argon (gemini-4-argon) Golden Remediations (argon_golden_remediations): Multi-file root-cause bug fixes staged through krusch-harness and verified against live unit test suites (node:test / pytest, exit code 0).
openai/gpt-oss-120b Elitist Code Audit Trajectories (elitist_120b_teacher): 11-phase AST-grounded codebase audits using krusch-git (search_symbols, dependency_graph, read_blob) with verbatim file:startLine-endLine citations, false-positive refutation (NOT FLAGGED), and Zero-Finding convergence (CLEAN_PASS).
Lego-First Idiom Synthesis & Rule-of-3 Anti-Bloat Gate (INV-LEGO-01..27): Assembling services from @krusch/toolkit v1.7 and krusch_toolkit v1.6 bricks (createApiServer, sseResponse, defineMcpTool, createPgPool, openSqliteDb, safeFetch, createRateLimiter), nominating >= 3-repo clusters via CANDIDATE_MEETS_RULE_OF_3, and rejecting < 3-repo utilities as BELOW_THRESHOLD_DO_NOT_EXTRACT.
KL-Anchored Stage-2 Preference Calibration (48 DPO Pairs, beta=0.1): Combines Referee CLEAN_PASS precision + P0/P1 recall pairs with action_over_prose, bounded_search_no_loop, and benign_execution_no_false_refusal tool-action pairs (100.0% preference accuracy).

⚡ Quickstart

1. Run Locally with Ollama (`Modelfile`, `adapter.gguf` & `~6.8s` GGUF Fusion Included)

git clone https://huggingface.co/krush420/qwen3-coder-30b-sre-lora
cd qwen3-coder-30b-sre-lora
ollama pull qwen3-coder-30b-moe:latest

# Fuse rank-16 attention LoRA into Q4_K/Q6_K base GGUF in ~6.8s (zero runtime GEMV overhead)
BASE_BLOB="$(ollama show --modelfile qwen3-coder-30b-moe:latest | awk '/^FROM / {print $2}')"
python3 fuse_lora_into_gguf.py --base-gguf "$BASE_BLOB" --adapter-gguf ./adapter.gguf --output-gguf ./qwen3-coder-30b-sre.gguf
ollama create qwen3-coder-30b-sre:latest -f Modelfile && rm -f ./qwen3-coder-30b-sre.gguf
ollama run qwen3-coder-30b-sre:latest

2. Transformers + PEFT

import torch
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base_id = "Qwen/Qwen3-Coder-30B-A3B-Instruct"
adapter_id = "krush420/qwen3-coder-30b-sre-lora"

tokenizer = AutoTokenizer.from_pretrained(base_id)
base_model = AutoModelForCausalLM.from_pretrained(
    base_id,
    dtype=torch.bfloat16,
    device_map="auto",
)
model = PeftModel.from_pretrained(base_model, adapter_id)

3. vLLM Multi-LoRA Serving

vllm serve Qwen/Qwen3-Coder-30B-A3B-Instruct \
  --enable-lora \
  --lora-modules code-auditor=krush420/qwen3-coder-30b-sre-lora

🔬 Wave-4 Unified Training Hyperparameters & Metrics

Hardware: NVIDIA RTX PRO 6000 Blackwell Server Edition (96 GB VRAM, sm_120; peak allocated VRAM 59.34 GB, 832.63s total wall time)
Base Model: Qwen/Qwen3-Coder-30B-A3B-Instruct (30.5B total / 3.3B active MoE parameters; 4-bit NF4 attention + frozen BF16 3D MoE experts)
LoRA Config: r=16, lora_alpha=32 (alpha/r = 2.0), lora_dropout=0.05, target_modules=["q_proj", "k_proj", "v_proj", "o_proj"] (13,369,344 trainable parameters across 48 layers; 51.03 MB GGUF v3 adapter)
Stage-1 Unified SFT (8192 ctx, 420 trajectories, lr=1.2e-4, 380.36s): Loss converged 1.7188 -> 1.4253 (train_loss=1.3884), token accuracy 67.00% -> 69.91% across native <tool_call> chains and code-audit trajectories
Stage-2 Unified DPO (48 balanced Referee + Tool-Action pairs x 3 epochs, beta=0.1, lr=1e-5, 253.49s): Loss 0.7014 -> 0.5238 (train_loss=0.5902), reward margin 0.000 -> +0.4031, 100.0% preference accuracy
Optimizer: paged_adamw_8bit, cosine LR schedule (warmup_ratio=0.05)

🏆 KruschStack Ecosystem Hermetic Benchmark Scorecard (`12` External Suites)

All per-query JSON audit logs and SHA-256 digests are published in krush420/sovereign-g4-benchmarks and interactive Pareto plots at spaces/krush420/sovereign-router-arena:

Pillar	Public Benchmark	Verified Sovereign Score
Agent Memory & Tool Execution	DolphinBench (`600` Tasks, `1.5M` Tokens, Official `gpt-5.6-sol`)	`42.67%` (`256/600`) on `30B` Wave-4 QLoRA+DPO (`+4` vs. `42.00%` Stock, `+11` on `Riley`; `85.73%` [`1,123/1,310`] mechanical checks, `$0.00` local inference, beats `Claude Sonnet 5 + Mem0/Honcho` & `MiniMax M3 + Hindsight`)
Security & Policy Floor	ETH SpyLab AgentDojo v1.2.2 (`1,046` episodes, `949` attacks)	`0.00%` Targeted ASR (`0/949`, `100%` Defense) · `69.07%` Benign (Spotlighting) & `83` hardline floor blocks
Code AST & Hybrid RRF	ACL 2025 CoIR (`2.39M` docs / `162K` Qs) & RepoBench-R (`47,998` Qs)	🥇 `#1` CoIR 10-Task `NDCG@10` (`56.40%`, `73.88%` `R@10` > `Voyage-Code-002` `56.26%`) · RepoBench-R `68.58%` `Acc@5`
SQL State & Deferred MCP	BIRD-SQL Mini-Dev (`500` Qs) & UC Berkeley BFCL v3 Live (`2,251` Qs)	`26.0% -> 40.2%` SQL EX (`1.55x`, `100%` valid SQL) · `59.9% -> 66.3%` BFCL Micro & `84.7%` Irrelevance at `-80.7%` schema tokens
Cascade Router (`L1–L3`)	Martian RouterBench (`73K`) · ACL 2026 LLMRouterBench (`12.5K`) · RouterArena (`8.4K`)	🥇 `#1` RouterBench ID/OOD `AIQ` (`-72.6%` cost @ `98%` `GPT-4`) · 🥇 `#1` LLMRouterBench Pareto (`0.4269`) · RouterArena CI `0.7883` (PR #169 open)
2PC Harness, Memory & Legal	Aider Polyglot (`225/225`) · ICLR 2026 BEAM (`2,000` Qs) · LegalBench-RAG (`6,889` Qs)	`19.56% -> 51.56%` `Pass@2` (`2.64x`, ties `Claude 3.5 Sonnet`) · `92.58%` `100K` & `93.00%` `10M` IE Recall · `26.74%` Legal Macro F1

Downloads last month: 88

GGUF

Model size

13.4M params

Architecture

qwen3moe

Hardware compatibility

We're not able to determine the quantization variants.

View all variants

Model tree for krush420/qwen3-coder-30b-sre-lora

Base model

Qwen/Qwen3-Coder-30B-A3B-Instruct

Adapter

(63)

this model

Datasets used to train krush420/qwen3-coder-30b-sre-lora

Space using krush420/qwen3-coder-30b-sre-lora 1

Evaluation results

Official gpt-5.6-sol Pass Rate (256/600 %) on DolphinBench (600 Tasks across 3 Personas / 1.5M Tokens, Official gpt-5.6-sol Judge)
test set Sovereign G4 & DolphinBench Hermetic Audit Logs

42.670
Deterministic Mechanical Tool Checks (1,123/1,310 %) on DolphinBench (600 Tasks across 3 Personas / 1.5M Tokens, Official gpt-5.6-sol Judge)
test set Sovereign G4 & DolphinBench Hermetic Audit Logs

85.730
Stage-2 KL-Anchored DPO Preference Accuracy (%) on Sovereign Code Audit, Lego-First & Agentic Tool-Calling SFT + DPO
Sovereign Code Audit SFT & DPO Distillation Pipeline

100.000
Seed Code-Audit Token Accuracy (%) on Sovereign Code Audit, Lego-First & Agentic Tool-Calling SFT + DPO
Sovereign Code Audit SFT & DPO Distillation Pipeline

93.360
100K Strict Primary Turn Recall (9-Cat %) on Official ICLR 2026 BEAM (100K-10M Splits, n=2,000)
Sovereign G4 Hermetic Audit Logs

92.580
10M Information Extraction Recall (%) on Official ICLR 2026 BEAM (100K-10M Splits, n=2,000)
Sovereign G4 Hermetic Audit Logs

93.000

Qwen3-Coder-30B-A3B — KruschStack Wave-4 Unified Agentic SRE & Tool-Caller LoRA (L2 Student)