Dicta-LM 3.0: Advancing The Frontier of Hebrew Sovereign LLMs
Dicta-LM 3.0 is a powerful open-weight collection of LLMs, trained on extensive corpora of Hebrew and English texts. The models are available for download and for unlimited use. The models set a new SOTA for their weight-class for Hebrew, both as base models and chat models.
This is our flagship model, a 24-billion-parameter reasoning model, with full precision (BF16), originally initialized from Mistral-Small-3.1-24B-Base-2503.
This model is a reasoning chat model, which means that before responding to any given message from the user, the model first thinks out the right way to respond in a designated thinking block.
π Try it out here: chat.dicta.org.il
For full details of this model please read our release blog post or the technical report.
You can view and access the full collection of base/instruct unquantized/quantized versions of DictaLM 3.0 here.
Instruction format
In order to leverage instruction fine-tuning, your prompt should be rendered using the chat template specified for this model. Most libraries deal with this automatically, so you can just let them do it.
Usage
We recommend using vLLM, but you can use Transformers as well:
Transformers
from transformers import pipeline
generator = pipeline('text-generation', model="dicta-il/DictaLM-3.0-24B-Thinking")
messages = [
{"role": "user", "content": "ΧΧΧΧ Χ¨ΧΧΧ ΧΧΧΧ Χ’ΧΧΧ?"},
{"role": "assistant", "content": "ΧΧΧ, ΧΧ Χ ΧΧ ΧΧΧΧ ΧΧΧ ΧΧΧ€ΧΧͺ ΧΧΧ₯ ΧΧΧΧΧ Χ‘ΧΧΧ ΧΧ¨Χ. ΧΧ ΧΧΧ‘ΧΧ£ ΧΧΧΧΧ§ ΧΧͺ ΧΧΧΧΧͺ ΧΧ ΧΧΧ Χ Χ©Χ ΧΧ’Χ ΧΧΧ¦ΧΧ₯ ΧΧΧ ΧΧ Χ©ΧΧ Χ ΧΧΧ©Χ ΧΧΧΧΧ!"},
{"role": "user", "content": "ΧΧΧ ΧΧ© ΧΧ ΧΧͺΧΧΧ ΧΧ ΧΧΧΧΧ Χ?"}
]
print(generator(messages)[0]['generated_text'][-1]) # just print the last message
# {'role': 'assistant', 'content': 'ΧΧΧ©ΧͺΧΧ© Χ©ΧΧΧ: "ΧΧΧ ΧΧ© ΧΧ ΧΧͺΧΧΧ ΧΧ ΧΧΧΧΧ Χ?" Χ’ΧΧΧ Χ ΧΧΧΧΧ Χ’Χ ΧΧͺΧΧΧ ΧΧΧΧΧ Χ. Χ‘Χ€Χ§Χ ΧΧͺΧΧΧ Χ§ΧΧΧ‘Χ, ΧΧΧΧ ΧΧ ΧΧ¨ΧΧΧ¦ΧΧΧͺ (ΧΧΧΧ Χ Χ€ΧΧΧΧ, ΧΧΧΧ Χ ΧΧΧ’ΧΧ Χ, ΧΧΧΧ Χ ΧΧ¨ΧΧ£, ΧΧΧ\'). Χ‘Χ€Χ§Χ ΧΧΧ¨ΧΧΧͺ Χ©ΧΧ ΧΧΧ¨ Χ©ΧΧ. Χ¦Χ¨ΧΧ ΧΧΧΧΧͺ ΧΧΧΧ ΧΧΧΧΧΧͺΧ.</think>ΧΧΧ! ΧΧΧΧ Χ ΧΧΧͺΧ ΧΧΧ ΧΧΧΧͺ Χ€Χ©ΧΧ ΧΧΧΧ Χ ΧΧΧ’Χ ΧΧ§ ΧΧΧ Χ©ΧΧΧΧ ΧΧΧΧ Χ’Χ ΧΧΧ¨ΧΧΧΧΧ, ΧΧΧ’Χ ΧΧΧΧ¨Χ§Χ. ΧΧΧΧ ΧΧΧͺΧΧΧ ΧΧ§ΧΧΧ‘Χ, ΧΧͺΧΧ‘Χ€Χͺ ΧΧΧ ΧΧ¨ΧΧΧ¦ΧΧΧͺ ΧΧΧ ΧΧͺ Χ©ΧͺΧΧΧΧ ΧΧΧͺΧ Χ‘ΧΧͺ ΧΧΧ.\n\n---\n\n## 1οΈβ£ ΧΧΧΧ Χ Χ§ΧΧΧ‘Χ (ΧΧΧΧ Χ ΧΧ‘ΧΧ‘Χ)\n\n| ΧΧ¨ΧΧΧ | ΧΧΧΧͺ |\n|------------|----------|\n| ΧΧΧΧΧ ΧΧΧ¦Χ ΧΧΧΧ (ΧΧ 2 ΧΧΧΧΧ ΧΧ ΧΧΧ ΧΧ ΧΧΧ) | 1 |\n| ΧΧ¨ΧΧ ΧΧΧ\'ΧΧ | 1 ΧΧ€ΧΧͺ |\n| ΧΧΧ₯ ΧΧΧΧΧ ΧΧ¨Χ (ΧΧ Χ'}
vLLM
vllm serve dicta-il/DictaLM-3.0-24B-Thinking --enable-auto-tool-choice --tool-call-parser hermes --reasoning_parser deepseek_r1
And then you can access it via the openai library:
from openai import OpenAI
client = OpenAI(
base_url="http://localhost:8000/v1",
api_key="sk-no-key-required"
)
response = client.chat.completions.create(
model="dicta-il/DictaLM-3.0-24B-Thinking",
messages=[
{"role": "user", "content": "Hello, how are you?"}
],
)
print(response.choices[0].message.content)
The reasoning traces should be available in the response structure in the designated fild.
The model supports tool-calling, enabling integration with external tools and APIs. For example how to use the tool calling, see the vLLM documentation.
Citation
If you use this model, please cite:
@article{Shmidman2025DictaLM3,
title={{Dicta-LM 3.0: Advancing The Frontier of Hebrew Sovereign LLMs}},
author={Shaltiel Shmidman and Avi Shmidman and Amir DN Cohen and Moshe Koppel},
year={2025},
publisher={{DICTA / Jerusalem, Israel}},
note={https://www.dicta.org.il/publications/DictaLM_3_0___Techincal_Report.pdf}
}
- Downloads last month
- 1,753
1-bit
2-bit
3-bit
4-bit
5-bit
6-bit
8-bit
16-bit
Model tree for unsloth/DictaLM-3.0-24B-Thinking-GGUF
Base model
dicta-il/DictaLM-3.0-24B-Thinking