kataguru/Industry-Coder-27B-v1.0-NVFP4 (Blackwell Native W4A4 FP4 + Vision + MTP)

Kataguru Industry-Coder-27B-v1.0-NVFP4 is the native NVIDIA Blackwell (sm_120, RTX 5090 / RTX PRO 6000) W4A4 FP4 production release of Industry-Coder-27B-v1.0 — a 27-billion-parameter dense flagship model engineered for industrial software engineering, algorithmic coding (LiveCodeBench v6), agentic tool calling (BFCL V4), industrial automation (PLC/CNC: Beckhoff TwinCAT 3 ST & Siemens SCL), European/Finnish patent engineering (EPO/PRH), and doctoral-level STEM reasoning (GPQA Diamond & GSM8K). All 80/80 shared weight_global_scale fusion groups are 100% unified, and all 333 vision tensors and 15 MTP speculative decoding tensors are preserved in full BF16 precision.

🇫🇮 Suomenkielinen dokumentaatio löytyy englanninkielisen osion alta.

🇬🇧 English Documentation

1. Summary of Kataguru Modifications

Field	Description / Value
Base model	`Qwen 3.8 27B` (64 layers: Gated DeltaNet + Full Attention + 333 Vision tensors + 15 MTP speculative tensors)
Kataguru modifications	4-cycle closed-loop Merge-and-Reset curriculum (`Cycle 1..4` + Tempering Elite Bank in Cycle 4 with strict non-decreasing difficulty), `r=16 qr_frozen_a` orthonormal subspace training, SOMA/ARA `0.00%` refusal alignment, Kataguru Quad-Mode Chat Template, and unified `weight_global_scale` Blackwell NVFP4 quantization
Language quality (EN & FI)	Native-level English and 100% flawless Finnish: natural, idiomatic Finnish across software engineering, industrial automation, patents, and STEM without morphological errors or calques
Quantization	Native Blackwell `nvfp4-pack-quantized` (`compressed-tensors`), with Vision (`333` tensors) and MTP (`15` tensors) preserved in full `BF16`
Weights size (Disk)	`28.55 GB` (`7` safetensors shards)
VRAM requirement (vLLM)	`2x RTX 5090 32GB (TP=2)` or `1x 32GB+ Blackwell GPU`
What was NOT modified	64-layer hybrid backbone architecture, 333 vision tensors, 15 MTP tensors, and 262,144-token native context window
Primary use cases	Production vLLM clusters: software engineering (Python, Rust, C, Java, JS), agentic `<tool_call>` workflows, PLC/CNC automation, patent claims, and STEM reasoning
Recommended runtime	`vllm/vllm-openai:v0.29.0-cu129` (`--tensor-parallel-size 2 --kv-cache-dtype fp8`)
Tested environment	2x NVIDIA GeForce RTX 5090 32GB Blackwell (`sm_120`, TP=2, CUDAGraphs enabled)

2. Official 15-Benchmark Evaluation (2x NVIDIA RTX 5090 Blackwell TP=2, NVFP4, CUDAGraphs)

All benchmarks were executed under identical production settings (2x RTX 5090 32GB TP=2, CUDAGraphs enabled, num_concurrent=16, FP8 KV cache):

#	Benchmark Task	Metric	Qwen 3.8 27B Base	Kataguru ChemBio-27B v2	Industry-Coder-27B-v1.0	Delta vs. Base
1	GSM8K Math (CoT 1024-tok)	Strict / Flex Exact Match	76.50 %	82.87 %	89.92 % (90.60 %)	+13.42 pp
2	HellaSwag	Linguistic Coherence (`acc_norm`)	82.10 %	83.48 %	82.74 %	+0.64 pp
3	MedQA	USMLE 4-options (`acc`)	78.40 %	81.23 %	82.56 %	+4.16 pp
4	BoolQ	Natural Language Logic (`acc`)	86.40 %	88.35 %	87.86 %	+1.46 pp
5	WinoGrande	Contextual Reasoning (`acc`)	75.80 %	77.43 %	76.24 %	+0.44 pp
6	TruthfulQA MC2	Multi-True Probability	54.80 %	55.90 %	53.47 %	-1.33 pp
7	TruthfulQA MC1	Single-True Factuality	36.10 %	37.82 %	34.76 %	-1.34 pp
8	PIQA	Physical Intuition (`acc_norm`)	81.20 %	81.61 %	81.77 %	+0.57 pp
9	OpenBookQA	Scientific Application (`acc_norm`)	44.20 %	45.00 %	45.60 %	+1.40 pp
10	ARC-Challenge	Advanced STEM Reasoning (`acc_norm`)	61.10 %	61.26 %	62.12 %	+1.02 pp
11	ARC-Easy	Foundational STEM (`acc_norm`)	81.30 %	82.66 %	81.99 %	+0.69 pp
$\Sigma_{11}$	11-Task Cognitive Average	Cognitive Composite	66.17 %	70.69 %	70.82 %	+4.65 pp
12	GPQA Diamond (198 PhD STEM)	Exact Match (`133/198`)	45.20 %	52.02 %	67.17 %	+21.97 pp
13	IFEval Strict	Prompt / Instruction Strict	48.20 % / 58.40 %	63.40 % / 74.10 %	70.43 % / 79.26 %	+22.23 % / +20.86 %
14	LiveCodeBench v6 (`175` tasks)	Pass@1 (Sandbox Execution)	26.86 %	32.00 %	36.57 % (64/175)	+9.71 pp
15	BFCL V4 (`Single-Turn`)	Overall AST (`Python 96.25%`, `Multiple 99.50%`)	62.60 %	65.58 %	71.78 %	+9.18 pp
$\Sigma_{15}$	15-Benchmark Composite	Official 15-Benchmark Mean	60.31 %	66.04 %	68.33 %	+8.02 pp

Breakdown of Coding & Tool-Calling Benchmarks

LiveCodeBench v6 (36.57% Pass@1 across 175 competitive programming tasks):
- Easy: 86.05% (37 / 43)
- Medium: 34.62% (18 / 52)
- Hard: 11.25% (9 / 80)
BFCL V4 Single-Turn (71.78% Overall, 85.49% Live Overall AST):
- Python Simple AST: 96.25% (Live 95.35%)
- Multiple AST (selecting the right tool from multiple candidates): 99.50% (Live 79.68%)
- JavaScript Simple AST: 80.00%
- Java Simple AST: 63.00%
- Irrelevance Detection (abstaining when no tool matches): 54.89% (Non-Live 61.25%)

3. Deployment with vLLM (2x RTX 5090 TP=2)

vllm serve kataguru/Industry-Coder-27B-v1.0-NVFP4 \
  --served-model-name industry-coder-27b \
  --tensor-parallel-size 2 \
  --gpu-memory-utilization 0.88 \
  --max-model-len 32768 \
  --quantization compressed-tensors \
  --kv-cache-dtype fp8 \
  --enable-chunked-prefill \
  --max-num-batched-tokens 8192 \
  --max-num-seqs 16 \
  --trust-remote-code \
  --enable-auto-tool-choice \
  --tool-call-parser qwen3_coder

4. Chat Template & Reasoning Modes

The model uses the Kataguru Quad-Mode Chat Template:

Default Mode (Direct Execution / Fast Tool Calling): Produces direct, concise engineering responses or <tool_call> invocations without unnecessary preamble.
Deep <think> Reasoning Mode: Activated via chat_template_kwargs={"enable_thinking": true} or by prefixing the prompt with {REASON:einstein} (10-angle synthesis) or {REASON:spoon} (7-step first-principles verification).

5. Known Limitations & Honest Assessment

Hardware Requirement: Native W4A4 FP4 execution requires NVIDIA Blackwell (sm_120 or newer) GPUs. For Ampere, Ada Lovelace, or Hopper GPUs, use kataguru/Industry-Coder-27B-v1.0-W4A16-AWQ or kataguru/Industry-Coder-27B-v1.0-GGUF.
BFCL Parallel Same-Function Calls: While single tool calls (Python Simple AST = 96.25%) and multi-tool selection (Multiple AST = 99.50%) achieve high accuracy, prompts that require invoking the exact same function multiple times in parallel within a single turn (Non-Live Parallel AST = 23.00%, Live Parallel AST = 43.75%) may emit only the first <tool_call> unless explicitly instructed to emit separate tool calls for each item.
TruthfulQA MC1/MC2 (34.76% / 53.47%): Scores ~1.3 pp below the base model on English urban-legend multiple-choice questions due to heavy specialization in STEM, code execution, and industrial engineering.

🇫🇮 Suomenkielinen Dokumentaatio

Kataguru Industry-Coder-27B-v1.0-NVFP4 on NVIDIA Blackwell (sm_120, RTX 5090 / RTX PRO 6000) -arkkitehtuurille kvantisoitu natiivi NVFP4-tuotantomalli (80/80 jaettua weight_global_scale-fuusioryhmää 100 % yhtenäisiä, 333 BF16-visiotensoria ja 15 BF16 MTP-spekulaatiotensoria).

1. Mitä Kataguru muutti? (Muutosyhteenveto)

Kenttä	Kuvaus / Arvo
Base model	`Qwen 3.8 27B` (64 kerrosta: Gated DeltaNet + Full Attention + 333 Visio-tensoria + 15 MTP-tensoria)
Katagurun muutos	4-syklinen Merge-and-Reset -kurrikulum (`Sykli 1..4` + Karkaisupankki Syklissä 4), `r=16 qr_frozen_a`, SOMA/ARA 0.00 % kieltäytyminen ja jaetun fuusioskaalan (`weight_global_scale`) Blackwell NVFP4 -kvantisointi
Suomen kielen laatu	100 % Virheetön suomi: Tuottaa virheetöntä suomenkielistä insinööri-, koodaus-, automaatio- ja patenttitekstiä
Kvantisointi	Natiivi Blackwell `nvfp4-pack-quantized` (`compressed-tensors`), Visio ja MTP täydellä BF16-tarkkuudella
Painojen koko (Levytila)	`28,55 Gt` (`7` safetensors-shardia)
VRAM-tarve (vLLM)	`2x RTX 5090 32GB (TP=2)` tai `1x 32GB+ Blackwell GPU`
Mitä EI muutettu	64-kerroksinen hybridirunko, 333 visiotensoria, 15 MTP-tensoria ja 262 144 tokenin konteksti-ikkuna
Ensisijainen käyttö	Tuotantotason vLLM-klusterit: ohjelmistotuotanto, agenttityökalut (`qwen3_coder`), PLC/CNC-automaatio, STEM ja suomenkielinen päättely
Suositeltu ajotapa	`vllm/vllm-openai:v0.29.0-cu129` (`--tensor-parallel-size 2 --kv-cache-dtype fp8`)
Testattu ympäristö	2x NVIDIA GeForce RTX 5090 32GB Blackwell (`sm_120`, TP=2, CUDAGraphs)

2. Tunnetut Rajoitukset

Vaatii NVIDIA Blackwell (sm_120 tai uudempi) -näytönohjaimen natiivia FP4-laskentaa varten (vanhemmilla Ampere/Ada-korteilla käytä kataguru/Industry-Coder-27B-v1.0-W4A16-AWQ- tai kataguru/Industry-Coder-27B-v1.0-GGUF-julkaisua).
BFCL Parallel AST (saman funktion toistokutsut yhdessä vuorossa): 23.00 % Non-Live / 43.75 % Live (Simple Python AST = 96.25 %, Multiple AST = 99.50 %).