eldiablo92/GPU_memory_sizer_for_LLM_inference

Front modif

Front modif

feat(front): improving documentation

feat(front): offloading front

feat(front): offloading front

feat(vLLM - KV cache - max context) : adding documentation and explainations

feat(vLLM): add quantization param + seperation between modele and KV for inference

feat(vLLM): add vllm simulation

feat(maxcontext + kvcache): Add Cuda overhead

feat(maxkvcache): improve display of information , add model quantiz option

feat(kvKachce): improve visualization add charts

feat(max-context-per-gpu): add GPU card models for simulation

feat(kv-cache): add model weight reduction per quantization level and total

feat: Add option to include model weights in max conctext compute + enhance front + fix when model do not fit + add models

feat: optim uv --no-sync

fix: CD pipeline

feat: makefile new commands + CD pipeline

fix: readme correction

fix: readme correction

fix: readme correction

fix: readme correction

fix: ci pipeline typo

feat: containerize app for Hugging Face Spaces

Preparing deployment : docker completed ready

Preparing deployment : docker

Reorg and cleaning

After ruff format

ci: add GitHub actions workflow

Initiating tests

cleaning dependencies

cleaning language : full english + correct typos

(feat) add screenshoot of the front

(feat) add screenshoot of the front

(feat) add screenshoot of the front

(feat) add screenshoot of the front

(reorg) Enhance Swagger + makefile

(reorg) refacto

(reorg) Add front fr + eng

(feat) Enhance readme style

(feat) Enhance CLI cmd file

(feat) Add readme

(feat) Add frontend

(feat) Add batch size

(feat) Add new models examples

(feat) Add new models examples

(reorg) fix all bugs

(reorg) moving modular

(feat) Add plot context vs memory plot

(feat) add seperation beteween model and kv cache

(fix) more details , bugs in convertion based on 1000 or 1024