tlm
github
x
fenn
the tiny language model. 135 million parameters on a Raspberry Pi Zero 2 W. runs locally.
compressed output
hardware
board
Pi Zero 2 W
cpu
ARM Cortex-A53
ram
512 MB
display
128x128 ST7735S
power
battery UPS
model
name
SmolLM2-135M-Instruct
quantization
Q4_K_M
size
101 MB
speed
~10 tok/s
response
16-22 sec
software
runtime
llama.cpp v1.0.0
binary
8 MB static
inference
tlm.py
display
lcd_display.py
os
Debian 13 aarch64