Instructions to use AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16 with Transformers:

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

pipe = pipeline("text-generation", model="AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16")
model = AutoModelForMultimodalLM.from_pretrained("AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16

SGLang

How to use AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16 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 "AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16" \
    --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": "AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16",
		"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 "AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16" \
        --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": "AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16 with Docker Model Runner:
```
docker model run hf.co/AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16
```

NOESIS / AMAImedia

Released as part of the NOESIS Professional Multilingual Dubbing Automation Platform (framework: DHCF-FNO — Deterministic Hybrid Control Framework for Frozen Neural Operators).

Founder: Ilia Bolotnikov
Organization: AMAImedia.com
X (Twitter): @AMAImediacom
LinkedIn: Ilia Bolotnikov
Telegram: @djbionicl
NOESIS version: v16.1
Release date: 2026-08-26

================================================================================ NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16 -- NOESIS Bundle README

✅ WIRED via Path A G4→G3 aligner v1 (2026-06-05). The plug-in incompatibility described below (R-SCENEMA-GEMMA4-PLUG-IN-INCOMPATIBLE) is now bridged by a trained non-linear aligner — see scenema_aligner_v1/ in this bundle. The aligner maps Gemma-4's output hidden states onto the Gemma-3 distribution that Scenema-DiT expects. Pipeline: Gemma-4 forward → aligner(hs_g4) → Scenema-DiT.

Quality (Path A v1, step 12000, full clean run RC=0): mean per-lang val_cos = 0.908; 12 of 13 languages ~0.92 (es,pt,de,it,ru,fr,en,sw,hi,zh ≈ 0.92-0.93; ja ≈ 0.910; ar ≈ 0.899); ko (Korean) = 0.728 — weak outlier, gate_passed=false is driven by Korean ONLY. Accepted as-is by operator 2026-06-05 (Korean fix deferred to a future korean-weighted retrain; the aligner is usable for the other 12 languages now).

The aligner is a NON-LINEAR module (LayerNorm → Linear-GELU-Linear + residual → per-layer scale/bias). It is applied to Gemma-4 OUTPUT hidden states at inference and CANNOT be weight-merged into Gemma-4's model.safetensors — folding a GELU/LayerNorm into transformer weights is mathematically impossible. The "merge" is a paired runtime: the Gemma-4 weights + the aligner sidecar, wired in sequence.

NOESIS provenance

Bundle : NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16 Parent bundle : NOESIS-3.5B-A0.5B-DUBBING-FILM / LTX2.3-5B-Scenema-Audio-DiT-BF16 Upstream : google/gemma-4-12b-it (Google LLC, Gemma Terms of Use) NOESIS variant : Verbatim upstream BF16 weights (NOT modified) + PAIRED Path A G4→G3 aligner v1 (scenema_aligner_v1/). NOESIS role : Wired Scenema text-encoder (G4 path). Path 1 of the three GEMMA4_INTEGRATION_HANDOFF_2026-06-04 paths — G4→G3 hidden-state aligner — was TRAINED and accepted: 1. ✅ G4→G3 aligner v1 (DONE 2026-06-05, mean 0.908, 12/13 langs ~0.92, ko=0.728 weak; trained on AWS L40S, full 12000 steps RC=0) 2. Retrain audio_aggregate_embed ($30-80) — not done 3. Route 2 v4 with G4 teacher ($20-50) — not done The aligner resolves the plain-swap incompatibility for 12 of 13 languages. Korean (ko=0.728) deferred. Downloaded : 2026-06-04 Last updated : 2026-08-29 (Path A G4→G3 aligner v1 trained + paired; status: held-for-experiments → WIRED)

================================================================================ ⚠️ License gate (Gemma Terms of Use)

Gemma 4 weights are released by Google under the Gemma Terms of Use (https://ai.google.dev/gemma/terms) and the Gemma Prohibited Use Policy (https://ai.google.dev/gemma/prohibited_use_policy):

Use is conditioned on accepting the Gemma Terms and complying with the Prohibited Use Policy.
Commercial use IS permitted under the Gemma Terms (no revenue cap) but Google's restrictions on use, redistribution, and derivative works still apply.
Redistribution must include a copy of the Gemma Terms and provide prominent attribution to Google as the upstream Licensor.
Any NOESIS-side fine-tune or adapter built on these weights becomes a Derivative Work subject to the same Terms.

This bundle is the unmodified upstream BF16 weights as published by Google. NOESIS has not modified the weights — see R-SCENEMA-GEMMA4-PLUG-IN-INCOMPATIBLE-WITHOUT-DIT-RETRAIN and the Route 2 experiment handoff for the planned downstream LoRA / adapter training paths.

================================================================================ Why this is held (not yet usable in production)

Per CLAUDE.md (UPD 2026-06-04 evening, R-SCENEMA-ROUTE2-ADAPTER-V3-RU-FAILS-LOCAL

Gemma 4 integration handoff):

Property	Gemma 3 12B (current teacher)	Gemma 4 12B (this bundle)
Shape: `[49 × 3840 → 188160]`	✅ matches	✅ matches
Per-layer hidden-state stacked-norm	~10,267,738	~23,464 (×437 smaller)
Per-channel hidden-state std	up to ~13	up to ~163,295 (×12,500 larger)
Plain plug-in into Scenema DiT	✅ works (R-SCENEMA-TEXT-ENCODER-PAIRED)	❌ fails — Russian phonemes only, no content words
`norm_and_concat_per_token_rms` alignment	n/a	partial — amplitude lost, alignment works on phonemes only

The shape compatibility is misleading: Scenema-DiT was trained on Gemma 3 hidden-state amplitude distributions. Gemma 4 substitution produces audio that contains real Russian phonemes but not the words of the input prompt. The fix requires either:

G4→G3 hidden-state alignment adapter (cheapest, recommended): train a small Linear / affine adapter that maps Gemma 4 last hidden layer onto the Gemma 3 distribution that Scenema-DiT expects. $5-15, A100 80GB, ~3-5h.
Retrain audio_aggregate_embed: regenerate the frozen projection in scenema-audio-pipeline.safetensors to expect Gemma 4 distributions natively. $30-80.
Route 2 v4 with Gemma 4 teacher: re-run the Route 2 cross-lingual KD with Gemma 4 as teacher (instead of Gemma 3). $20-50.

All three paths are CLOUD-SIDE experiments — none fit on the RTX 3060 6 GB local node.

================================================================================ Architecture

Property	Value
Architecture	Gemma 4 (12B instruction-tuned) — Google's next-gen Gemma
Total parameters	~12B
File size	~23.9 GB (single shard `model.safetensors`)
Precision	BF16 (verbatim upstream)
Hidden-state shape	`[seq, 3840]` (same as Gemma 3 12B)
Hidden-state distribution	DIFFERENT from Gemma 3 (see table above)
Tokenizer	Gemma 4 tokenizer
License	Gemma Terms of Use

================================================================================ Bundle Contents

.
├── README.md
├── LICENSE                       # Gemma Terms of Use + NOESIS notice
├── config.json
├── generation_config.json
├── processor_config.json
├── chat_template.jinja
├── tokenizer.json
├── tokenizer_config.json
├── model.safetensors             # BF16 weights (~23.9 GB single shard)
└── scenema_aligner_v1/           # Path A G4→G3 aligner v1 (PAIRED runtime)
    ├── aligner_g4_to_g3_v1.pt    # 253 MB — non-linear aligner weights
    ├── apply_aligner.py          # load + apply helper (inference wrapper)
    ├── train_pathA_aligner.py    # GemmaFeatureAligner class + trainer
    ├── metrics_final.json        # per-lang val_cos (mean 0.908, ko 0.728)
    └── step.json                 # training state (step 12000)

================================================================================ Usage — Gemma-4 → aligner → Scenema-DiT

# from inside scenema_aligner_v1/
from apply_aligner import load_aligner
aligner = load_aligner("aligner_g4_to_g3_v1.pt", device="cuda")

# hs_g4 = Gemma-4-12B last-hidden-state stack [B, T, 3840]
hs_aligned = aligner(hs_g4)          # mapped onto Gemma-3 distribution
# feed hs_aligned into Scenema-DiT cross-attention as the text condition

The aligner is applied at INFERENCE between Gemma-4's forward pass and the Scenema-DiT — it is NOT merged into model.safetensors (non-linear, cannot be folded into transformer weights). Korean (ko) output is weak (0.728); the other 12 languages are production-usable (~0.92).

================================================================================ NOESIS Scenema Family

Bundle	Role
Gemma3-12B-it-Qat-Q4_0-Unquantized-BF16	Production KD teacher (Gemma 3, 3840-d, fully plug-compatible)
NOESIS-Gemma3-4B-Scenema-LTX2.3-BF16	Primary student (cinema + autoif LoRA merged)
NOESIS-Gemma3-1B-BF16	Recipe-validation student (smoke tests)
NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16 ← this bundle	WIRED via Path A G4→G3 aligner v1 (12/13 langs ~0.92, ko weak)

================================================================================ NOESIS Sealed Rules

R-SCENEMA-GEMMA4-ALIGNER-V1-WIRED-2026-06-05 (sealed 2026-06-05, SUPERSEDES the "plug-in incompatible" status) The G4→G3 hidden-state aligner (Path 1) was TRAINED and accepted: scenema_aligner_v1/aligner_g4_to_g3_v1.pt (253 MB, non-linear MLP + per-layer scale/bias). Full 12000-step run on AWS L40S, RC=0. Results: mean per-lang val_cos 0.908; 12/13 langs ~0.92; ko (Korean) = 0.728 (weak — gate_passed=false driven by Korean only). Operator accepted as-is 2026-06-05; Korean fix deferred to a future korean-weighted retrain. The aligner is applied at inference (Gemma-4 forward → aligner → Scenema-DiT) and is NOT weight-merged into model.safetensors (non-linear, cannot fold into transformer weights). This resolves the previous plain-swap incompatibility for 12 of 13 languages.

R-SCENEMA-GEMMA4-PLUG-IN-INCOMPATIBLE-WITHOUT-DIT-RETRAIN (sealed 2026-06-04 — HISTORICAL, now bridged by aligner v1 above) Gemma 4 12B CANNOT PLAIN-swap into Scenema-DiT (shape matches but hidden-state distribution differs ×437 stacked-norm / ×12,500 std). Bridged by R-SCENEMA-GEMMA4-ALIGNER-V1-WIRED-2026-06-05 (aligner v1).

R-GEMMA-TERMS-PRESERVE Upstream Gemma Terms of Use preserved verbatim in LICENSE. Bound by Google's Gemma Terms + Prohibited Use Policy.

R-VENDORED-INTERNAL Internal vendor copy inside parent NOESIS-3.5B-A0.5B-DUBBING-FILM / LTX2.3-5B-Scenema-Audio-DiT-BF16 bundle.

R-HELD-FOR-EXPERIMENTS Held on disk as the substrate for cloud-side Route 2 G4→G3 adapter experiments. Not yet a NOESIS LoRA fine-tune. NOESIS-side LoRA / adapter training will happen on cloud (out of scope for RTX 3060 6 GB local node).

R-DUBBING-FILM-SCOPE (sealed 2026-04-29) NOESIS = professional audio dubbing. This bundle is research / experiment substrate for the Scenema text-encoder upgrade path.

R-NOESIS-FINAL-ARTIFACT-PATHS (sealed 2026-05-27) Canonical path: models/llm/NOESIS-3.5B-A0.5B-DUBBING-FILM/ LTX2.3-5B-Scenema-Audio-DiT-BF16/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16/

R-NEVER-DELETE-WITHOUT-EXPLICIT-CONSENT (sealed 2026-05-21) MUST NOT be deleted without explicit operator instruction "удали

================================================================================ Handoff documents

Per CLAUDE.md banner 2026-06-04:

docs/GEMMA4_INTEGRATION_HANDOFF_2026-06-04.md — 3 experiment paths
docs/ROUTE2_V4_FEEDBACK_TO_CLOUD_2026-06-04.md — Route 2 v4 results

Related sealed rules:

R-SCENEMA-ROUTE2-ADAPTER-V3-RU-FAILS-LOCAL — adapter v3 (val_cos 0.9824) NOT producing Russian locally; en-overfit. Wait for adapter v5 (full 12k steps + balanced 13 langs + per-lang val_cos ≥ 0.95).
R-SCENEMA-AUTHENTIC-LTX2-NOT-DRAMABOX — authentic LTX-2 packages only, NOT DramaBox.
R-SCENEMA-STANDARD-GEMMA-NOT-SIKAWORLD — standard Lightricks Gemma (Gemma 3), NOT third-party variants.

================================================================================ Upstream Citation

Upstream: https://huggingface.co/google/gemma-4-12b-it Gemma Terms: https://ai.google.dev/gemma/terms Prohibited Use: https://ai.google.dev/gemma/prohibited_use_policy

================================================================================

NOESIS — Deterministic Hybrid Control Framework for Frozen Neural Operators (DHCF-FNO). Copyright (c) 2026 AMAImedia.com. All rights reserved. Gemma 4 12B weights © 2025-2026 Google LLC, released under the Gemma Terms of Use (https://ai.google.dev/gemma/terms).

Downloads last month: 814

Safetensors

Model size

12B params

Tensor type

BF16

Collection including AMAImedia/NOESIS-Gemma4-12B-it-Qat-Q4_0-Unquantized-BF16

NOESIS-Original-Trained-Models

Collection

24 items • Updated 12 days ago