CoolFace
Modelpublic

sathishphdai/network-engineer-slm-5m

sourceHugging Facemitupdated 7mo agoView on Hugging Face
0likes15downloads
Model Card

Network Engineer-SLM: Role-Based Small Language Model

A LLaMA-style transformer (~1019.5M params, ~1.02B) trained from scratch for the Network Engineer role. Supports up to 5M token context via RoPE with gradient checkpointing.

Architecture

ComponentValue
ArchitectureLLaMA-style (RoPE + RMSNorm + SwiGLU)
Parameters~1019.5M (~1.02B)
Layers32
Heads20
Embedding1600
Max Context5,000,000 tokens
Max Output5,000,000 tokens
Vocab20,659 BPE
Model Size~4 GB (fp32)

Training

  • —Best eval loss: 5.257461833953857
  • —Trained with gradient checkpointing on Apple M4 (MPS)
  • —3 epochs, batchsize=1, gradaccum=16

Usage

python
from huggingface_hub import hf_hub_download
from tokenizers import Tokenizer

model_path = hf_hub_download("sathishphdai/network-engineer-slm-5m", "model.safetensors")
tokenizer_path = hf_hub_download("sathishphdai/network-engineer-slm-5m", "network_engineer_tokenizer.json")
tokenizer = Tokenizer.from_file(tokenizer_path)