โšก EvoNet-3B-Reasoning (Edge AI Edition)

EvoNet-3B-Reasoning lร  phiรชn bแบฃn rรบt gแปn (Lightweight) thuแป™c hแป‡ sinh thรกi EvoNet Security Audit System. ฤฦฐแปฃc xรขy dแปฑng trรชn nแปn tแบฃng Llama-3.2-3B-Instruct mแป›i nhแบฅt cแปงa Meta vร  tรญch hแปฃp tฦฐ duy suy luแบญn logic, mรด hรฌnh nร y ฤ‘ฦฐแปฃc thiแบฟt kแบฟ ฤ‘แบทc biแป‡t cho cรกc thiแบฟt bแป‹ cรณ phแบงn cแปฉng giแป›i hแบกn (Edge Devices, Mobile, Low-VRAM GPUs).

Dรน chแป‰ cรณ kรญch thฦฐแป›c 3 tแปท tham sแป‘, mรด hรฌnh vแบซn giแปฏ ฤ‘ฦฐแปฃc khแบฃ nฤƒng phรขn tรญch bแบฃo mแบญt tแปซng bฦฐแป›c (Step-by-step reasoning) vร  giao tiแบฟp tiแบฟng Viแป‡t trรดi chแบฃy.

๐Ÿš€ ฤiแปƒm nแป•i bแบญt

  • Siรชu nhแบน & Siรชu tแป‘c: Vแป›i kรญch thฦฐแป›c cแปฑc nhแป, cรณ thแปƒ chแบกy mฦฐแปฃt mร  trรชn cรกc mรกy tรญnh cรก nhรขn khรดng cรณ GPU chuyรชn dแปฅng, hoแบทc tรญch hแปฃp trแปฑc tiแบฟp vร o แปฉng dแปฅng di ฤ‘แป™ng.
  • Tฦฐ duy phรขn tรญch (Reasoning): Tรญch hแปฃp LoRA Reasoning giรบp mรด hรฌnh suy nghฤฉ trฦฐแป›c khi trแบฃ lแปi, cแบฃi thiแป‡n ฤ‘แป™ chรญnh xรกc trong cรกc bร i toรกn nhแบญn diแป‡n mรฃ ฤ‘แป™c.
  • Hแป— trแปฃ Tiแบฟng Viแป‡t: Khแบฃ nฤƒng sinh ngแปฏ cแบฃnh vร  giแบฃi thรญch lแป—i bแบฃo mแบญt bแบฑng tiแบฟng Viแป‡t tแปฑ nhiรชn.

โš ๏ธ Giแป›i hแบกn (Limitation Notice)

Do giแป›i hแบกn vแบญt lรฝ vแป sแป‘ lฦฐแปฃng tham sแป‘ (3B so vแป›i 8B), mรด hรฌnh cรณ thแปƒ gแบทp hiแป‡n tฦฐแปฃng Hallucination (แบฃo giรกc) trong cรกc kแป‹ch bแบฃn phรขn tรญch lแป— hแป•ng quรก sรขu hoแบทc giแบฃi thรญch sai cฦก chแบฟ hoแบกt ฤ‘แป™ng cแปงa mแป™t sแป‘ payload phแปฉc tแบกp. ๐Ÿ‘‰ Khuyแบฟn nghแป‹: Sแปญ dแปฅng phiรชn bแบฃn EvoNet-8B-Reasoning cho cรกc mรดi trฦฐแปng Production ฤ‘รฒi hแปi ฤ‘แป™ chรญnh xรกc tuyแป‡t ฤ‘แป‘i.

๐Ÿ› ๏ธ Chi tiแบฟt kแปน thuแบญt

  • Kiแบฟn trรบc: Llama 3.2
  • Kรญch thฦฐแป›c: 3 Billion Parameters
  • Base Model: unsloth/Llama-3.2-3B-Instruct
  • Ngฦฐแปi phรกt triแปƒn: Phong Huแปณnh (EvoNet)

๐Ÿ’ป Hฦฐแป›ng dแบซn sแปญ dแปฅng (Python)

Mรด hรฌnh cรณ thแปƒ chแบกy dแป… dร ng trรชn bแบฅt kแปณ GPU nร o (thแบญm chรญ lร  CPU) vแป›i thฦฐ viแป‡n transformers.

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "EvoNet/EvoNet-3B-Reasoning"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.float16,
    device_map="auto" # Chแบกy siรชu mฦฐแปฃt trรชn cรกc GPU nhแป
)

messages = [
    {"role": "system", "content": "Bแบกn lร  EvoNet Pentester. Hรฃy suy nghฤฉ tแปซng bฦฐแป›c trฦฐแป›c khi trแบฃ lแปi."},
    {"role": "user", "content": "Phรขn tรญch payload sau: `admin' AND (SELECT 1 FROM (SELECT SLEEP(5))A) AND '1'='1`. ฤรขy lร  lแป—i gรฌ?"}
]

prompt_text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(prompt_text, return_tensors="pt").to(model.device)

outputs = model.generate(**inputs, max_new_tokens=1024, temperature=0.7)
print(tokenizer.decode(outputs[0][inputs['input_ids'].shape[-1]:], skip_special_tokens=True))
Downloads last month
9
Safetensors
Model size
3B params
Tensor type
F16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for EvoNet/EvoNet-3B-Reasoning

Finetuned
(615)
this model
Quantizations
3 models