Instructions to use AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16 with Transformers:

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

pipe = pipeline("text-generation", model="AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-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/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16

SGLang

How to use AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-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/Emo-1T-14B-A1B-OLMoE-NOESIS-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/Emo-1T-14B-A1B-OLMoE-NOESIS-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/Emo-1T-14B-A1B-OLMoE-NOESIS-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/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16 with Docker Model Runner:
```
docker model run hf.co/AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16
```

Emo-1T-14B-A1B-OLMoE-NOESIS-BF16

BF16 dtype-repack of allenai/Emo_1b14b_1T — original FP32 floating-point weights losslessly cast to bfloat16 for LoRA / DoRA / PEFT compatibility and reduced disk footprint. The model architecture, parameter values, tokenizer, and configuration are identical to upstream — only the IEEE-754 storage dtype was changed.

License preserved end-to-end — see LICENSE in this repo for the full text and attribution chain.

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: @AMAImediacom
NOESIS version: v15.8
Repack date: 2026-05-19

Summary

Mixture-of-Experts language model trained with the Expert Modularity Objective (EMO) — documents constrained to route through shared expert pools during training, enabling domain-specific expert subsets to be deployed in isolation. Pretrained on 1T tokens (OLMoE pretraining mix) + 50B token annealing under the EMO objective.

Use case inside NOESIS

Research model for emergent modularity / domain-specific expert routing. English-only. Inside NOESIS this is out-of-scope for the dubbing pipeline (single-language, research-oriented) — kept as research reference.

What changed vs upstream

Aspect	Upstream	This bundle
Floating-point storage dtype	FP32	bfloat16
`config.json` `torch_dtype`	as-is	`bfloat16`
`model.safetensors.index.json` `total_size`	as-is	recomputed
Tokenizer / chat template / modeling code	as-is	unchanged
Number of parameters	as-is	unchanged
Value-level transformation beyond dtype cast	—	none
Disk size	51 GB	26 GB

Architecture

Property	Value
Immediate parent	`allenai/Emo_1b14b_1T`
Architecture	`EmoForCausalLM`
Architecture base / lineage	OLMoE-family MoE (128 experts = 127 routed + 1 shared, k=8 active per token, Expert-Modularity-Objective pretraining)
Parameters	1B active / 14B total
Hidden size	2048
Num hidden layers	16
Attention heads / KV heads	16
Vocab size	100352
Max position embeddings	4096
Format	bfloat16
Bundle size on disk	26 GB
License	Other (see upstream — typically Apache 2.0 OLMoE family policy)
Project page	https://huggingface.co/allenai/Emo_1b14b_1T
Paper / arxiv	arxiv:2605.06663

Repack tooling

CPU-only sharded repack via repack_fp32_to_bf16.py — reads each shard with safetensors.safe_open, casts floating-point tensors to torch.bfloat16, rewrites the shard, updates the index manifest. No GPU involvement, no value-level transformation beyond the IEEE-754 dtype cast.

Performance reference (RTX 3060 laptop, NVMe SSD):

Single 5 GB FP32 shard cast → ~28-40 sec
Full 51 GB → 26 GB in 1 pass, sharded

Use cases (for the BF16 bundle)

✅ LoRA / DoRA / IA³ fine-tuning that requires a dtype=torch.bfloat16 base
✅ Bitsandbytes NF4 / AWQ-INT4 / GPTQ quantization (these tools prefer BF16 input)
✅ Inference on Ampere+ / MI200+ hardware with native BF16 support
✅ KD-teacher (forward-only) where BF16 storage saves bandwidth
❌ Full-parameter fine-tuning of weights — use FP32/BF16 master weights pattern; storage dtype alone is insufficient

Quick start

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

repo = "AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16"

tokenizer = AutoTokenizer.from_pretrained(repo, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    repo,
    dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True,
).eval()

Sealed rules (NOESIS DHCF-FNO)

R-DTYPE-REPACK-BF16 — pure IEEE-754 dtype cast from FP32 to bfloat16. No value-level transformation, no LoRA merge, no architectural change. Equivalent to loading upstream with dtype=torch.bfloat16 and saving, but materialised on disk.
R-UPSTREAM-OTHER — upstream Other (see upstream — typically Apache 2.0 OLMoE family policy) preserved end-to-end via the LICENSE file in this repo. AMAImedia adds only a derivative-work notice for the repack step.
R-NO-VALUE-TRANSFORM — no fine-tuning, no distillation, no merge has been applied between upstream and this repo. Outputs are bit-for-bit equivalent up to the precision difference of the dtype cast.

License & attribution

This bundle inherits Other (see upstream — typically Apache 2.0 OLMoE family policy) from allenai/Emo_1b14b_1T. Original model card, citation, and attribution from upstream apply without modification. See LICENSE in this repo for the complete text plus the NOESIS derivative-work NOTICE.

Citation

@misc{noesis2026emo1t14ba1bolmoenoesisbf16bf16,
  title  = {NOESIS DHCF-FNO :: Emo-1T-14B-A1B-OLMoE-NOESIS-BF16 — BF16 dtype-repack derivative},
  author = {Bolotnikov, Ilia and AMAImedia},
  year   = {2026},
  note   = {BF16 dtype-repack of allenai/Emo_1b14b_1T for LoRA / PEFT
            compatibility. 26 GB on disk, Other (see upstream — typically Apache 2.0 OLMoE family policy)
            preserved end-to-end.},
  url    = {https://huggingface.co/AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16}
}

Please also cite the upstream model when using this bundle. See the upstream README and LICENSE in this repo for citation requirements.

Author

Founder: Ilia Bolotnikov
Organization: AMAImedia.com
X (Twitter): @AMAImediacom
LinkedIn: Ilia Bolotnikov
Telegram: @AMAImediacom
NOESIS version: v15.8
Repack date: 2026-05-19
HF repo: AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16
Upstream: allenai/Emo_1b14b_1T

Produced 2026-05-19 by NOESIS DHCF-FNO v15.8 — AMAImedia.com

Downloads last month: 27

Safetensors

Model size

14B params

Tensor type

BF16

Model tree for AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16

Base model

allenai/Emo_1b14b_1T

Finetuned

(1)

this model

Paper for AMAImedia/Emo-1T-14B-A1B-OLMoE-NOESIS-BF16

EMO: Pretraining Mixture of Experts for Emergent Modularity

Paper • 2605.06663 • Published 17 days ago • 12