Mungert/TildeOpen-30b-GGUF

TildeOpen-30b GGUF Models

Model Generation Details

This model was generated using llama.cpp at commit 408ff524.

---

Quantization Beyond the IMatrix

I've been experimenting with a new quantization approach that selectively elevates the precision of key layers beyond what the default IMatrix configuration provides.

In my testing, standard IMatrix quantization underperforms at lower bit depths, especially with Mixture of Experts (MoE) models. To address this, I'm using the --tensor-type option in llama.cpp to manually "bump" important layers to higher precision. You can see the implementation here:
👉 Layer bumping with llama.cpp

While this does increase model file size, it significantly improves precision for a given quantization level.

I'd love your feedback—have you tried this? How does it perform for you?

---

Click here to get info on choosing the right GGUF model format

---

Developed by: Tilde.ai
Funded by: European Commission via EuroHPC JU Large AI Grand Challenge
Model type: A 30B parameter dense decoder-only transformer
Languages: Albanian, Bosnian, Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Hungarian, Icelandic, Irish, Italian, Latgalian, Latvian, Lithuanian, Macedonian, Maltese, Montenegrin, Norwegian, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Slovene, Spanish, Swedish, Turkish, Ukrainian as well of mathematical proofs, programming code and XML documents containing translation data
License: CC-BY-4.0

Mission statement

TildeOpen LLM is an open-source foundational language model built to serve underrepresented Nordic and Eastern European languages. Developed with European Commission funding and trained on the LUMI supercomputer, this 30B+ parameter model addresses the performance gaps that speakers of 19 focus languages—representing over 165 million people—face with existing AI systems.
The model employs an equitable tokeniser and curriculum-learning approach to ensure fair representation across less-resourced languages, moving beyond the typical English-centric design of most language models. As an open-source project, TildeOpen LLM enables transparent research and community-driven development while maintaining European technological independence.
This foundational model is not yet adapted to follow instructions or aligned with safety features. The next version being built on top of this model will be a specialised translation model, leveraging TildeOpen LLM's multilingual foundation to provide high-quality translation capabilities across the supported European language pairs.

Model training details

We train TildeOpen LLM using the Tilde's branch of EleutherAI's open-source GPT-NeoX framework on LUMI supercomputer's 768 AMD MI250X GPUs. The foundational model training involves 450,000 updates with a constant batch size of 4,718,592 tokens, using a constant learning rate followed by a cooldown phase across 2 trillion tokens. Training consists of three distinct data sampling phases. First, all languages are sampled uniformly to ensure equal representation. Second, languages are sampled according to their natural distribution to ensure that the model sees as much data from languages with larger speaker bases as possible. Finally, we return to uniform sampling across all languages. This three-phase approach ensures TildeOpen LLM develops balanced multilingual capabilities while maintaining strong performance across all target languages, particularly the underrepresented European languages.

Model Hyper-Parameters

Parameter	Value
Sequence Length	8192
Number of Layers	60
Embedding Size	6144
FFN Hidden Size	21504
Number of Heads	48
Number of KV Heads (GQA)	8
Activation Function	SwiGLU
Position Encodings	RoPE
Layer Norm	RMSNorm
Embedding Parameters	8.05E+08
LM Head Parameters	8.05E+08
Non-embedding Parameters	2.91E+10
Total Parameters	3.07E+10

Tokeniser details

We built the TildeOpen LLM tokeniser to ensure equitable language representation across languages. Technically, we trained the tokeniser to represent the same text regardless of the language it is written in, using a similar number of tokens. In practice, TildeOpen LLM will be more efficient and faster than other models for our focus languages, as writing out answers will require fewer steps. For more details on how TildeOpen LLM compares against other models, see TILDE Bench!

Running model using HF transformers

When loading the tokeniser, you must set use_fast=False.

from transformers import AutoTokenizer, AutoModelForCausalLM

# Load tokenizer + model
tokenizer = AutoTokenizer.from_pretrained("TildeAI/TildeOpen-30b", use_fast=False)
model = AutoModelForCausalLM.from_pretrained(
    "TildeAI/TildeOpen-30b",
    torch_dtype=torch.bfloat16,
    device_map="auto"
)

# Tokenize
inputs = tokenizer(user_in, return_tensors="pt").to(model.device)

# Generate (greedy, deterministic)
outputs = model.generate(
    **inputs,
    max_new_tokens=512,
    repetition_penalty=1.2,
    do_sample=False,
)

Evaluation

Per-Character Perplexity

What is Perplexity? Perplexity measures how well a language model predicts text. A model with low perplexity makes accurate predictions consistently, while a high perplexity means the model is frequently "surprised" by unexpected words or patterns. Lower perplexity indicates the model has learned language patterns more effectively. It's less "surprised" by what it encounters because it better understands how the language works. Perplexity fairly evaluates how well each model handles:

• Spelling accuracy across a diverse vocabulary
• Grammar rules that span multiple words
• Sentence structure and flow
• Language-specific patterns (how different languages form plural forms or compound words)

Why Character-Level? Different language models use different internal vocabularies - some break text into whole words, others into word fragments, and some into individual characters. This makes direct comparison difficult. Character-level perplexity creates a standardised comparison by calculating how well each model would theoretically perform if we measured their predictions character-by-character. We're not changing how the models work - instead, we use mathematical conversion to approximate their character-level performance based on their predictions.

Why does this Matter? Models with lower perplexity generally perform better on real-world tasks like text generation, translation, and understanding context. It's a reliable indicator of overall language competency across different applications.

What data did we use? We use WMT24++ as it is a multilingual, language-parallel evaluation set that none of the models have seen during training. WMT24++ is a composite of texts from news, literature, speech, and social media; thus, it is suitable for foundational model benchmarking.

Language	TildeOpen 30b	Gemma 2 27b	EuroLLM 22B Prev.	ALIA 40B
Bulgarian	2.0539	2.2184	2.1985	2.1336
Czech	2.1579	2.3522	2.3221	2.2719
Danish	2.003	2.1517	2.1353	2.0805
German	1.8769	1.9285	1.9452	1.904
English	2.0378	1.9525	2.0568	2.0261
Spanish	1.9503	1.9752	2.0145	1.9369
Estonian	2.1711	2.5747	2.3852	2.325
Finnish	2.0497	2.288	2.2388	2.1831
French	1.8978	1.9355	1.9282	1.9084
Croatian	2.1147	2.544	2.4905	2.2433
Hungarian	2.0539	2.2228	2.2256	2.1635
Icelandic	2.0873	3.0329	4.7908	3.957
Italian	1.9565	2.0137	2.0098	1.9887
Lithuanian	2.1247	2.4175	2.3137	2.3075
Latvian	2.1439	2.5355	2.3141	2.3276
Dutch	1.9333	2.0312	2.0079	1.9904
Norwegian	2.1284	2.2862	2.3506	2.2253
Polish	2.0241	2.1294	2.0803	2.0803
Portuguese	1.9899	2.0597	2.0272	2.0187
Romanian	2.0196	2.1606	2.1641	2.1114
Russian	2.0424	2.09	2.1095	2.0871
Slovak	2.1192	2.338	2.3029	2.2609
Slovenian	2.1556	2.4443	2.3398	2.2589
Serbian	2.2469	2.6351	4.2471	2.3743
Swedish	2.041	2.1809	2.1464	2.1211
Turkish	2.0997	2.247	2.2202	2.232
Ukrainian	2.1376	2.2665	2.2691	2.2086

---

🚀 If you find these models useful

Help me test my AI-Powered Quantum Network Monitor Assistant with quantum-ready security checks:

👉 Quantum Network Monitor

The full Open Source Code for the Quantum Network Monitor Service available at my github repos ( repos with NetworkMonitor in the name) : Source Code Quantum Network Monitor. You will also find the code I use to quantize the models if you want to do it yourself GGUFModelBuilder

💬 How to test:
Choose an AI assistant type:

• TurboLLM (GPT-4.1-mini)
• HugLLM (Hugginface Open-source models)
• TestLLM (Experimental CPU-only)

What I’m Testing

I’m pushing the limits of small open-source models for AI network monitoring, specifically:

• Function calling against live network services
• How small can a model go while still handling:
  • Automated Nmap security scans
  • Quantum-readiness checks
  • Network Monitoring tasks

🟡 TestLLM – Current experimental model (llama.cpp on 2 CPU threads on huggingface docker space):

• ✅ Zero-configuration setup
• ⏳ 30s load time (slow inference but no API costs) . No token limited as the cost is low.
• 🔧 Help wanted! If you’re into edge-device AI, let’s collaborate!

Other Assistants

🟢 TurboLLM – Uses gpt-4.1-mini :

• **It performs very well but unfortunatly OpenAI charges per token. For this reason tokens usage is limited.
• Create custom cmd processors to run .net code on Quantum Network Monitor Agents
• Real-time network diagnostics and monitoring
• Security Audits
• Penetration testing (Nmap/Metasploit)

🔵 HugLLM – Latest Open-source models:

• 🌐 Runs on Hugging Face Inference API. Performs pretty well using the lastest models hosted on Novita.

💡 Example commands you could test:

1. "Give me info on my websites SSL certificate"
2. "Check if my server is using quantum safe encyption for communication"
3. "Run a comprehensive security audit on my server"
4. '"Create a cmd processor to .. (what ever you want)" Note you need to install a Quantum Network Monitor Agent to run the .net code on. This is a very flexible and powerful feature. Use with caution!

Final Word

I fund the servers used to create these model files, run the Quantum Network Monitor service, and pay for inference from Novita and OpenAI—all out of my own pocket. All the code behind the model creation and the Quantum Network Monitor project is open source. Feel free to use whatever you find helpful.

If you appreciate the work, please consider buying me a coffee ☕. Your support helps cover service costs and allows me to raise token limits for everyone.

I'm also open to job opportunities or sponsorship.

Thank you! 😊