Update README.md

README.md

@@ -72,7 +72,6 @@ The model was trained on a synthetic `.jsonl` dataset (`fridgebuddy_dataset_en.j
 ### Training Procedure
 
 - LoRA fine-tuning with Unsloth + PEFT
-- Training done on Google Colab (Tesla T4)
 - Float32 precision
 - 3 epochs — 36 steps
 
@@ -88,14 +87,6 @@ The model was trained on a synthetic `.jsonl` dataset (`fridgebuddy_dataset_en.j
 
 The model was qualitatively evaluated using generation on unseen prompts. It responded well to basic kitchen questions, but showed expected limitations in vocabulary and instruction-following depth.
 
-## Environmental Impact
-
-- **Hardware Type:** Tesla T4
-- **Hours used:** ~5 minutes (100 samples × 3 epochs)
-- **Cloud Provider:** Google Colab
-- **Compute Region:** Not specified
-- **Carbon Emitted:** Negligible
-
 ## Technical Specifications
 
 ### Model Architecture and Objective
@@ -103,11 +94,9 @@ The model was qualitatively evaluated using generation on unseen prompts. It res
 - Base: `gemma-3-1b-it`, 1.3B parameters
 - Finetuning: LoRA adapters via PEFT
 
-### Compute Infrastructure
-
-- **Hardware:** Google Colab (GPU: Tesla T4)
-- **Software:** PEFT 0.15.2, Transformers, Accelerate, Unsloth
-
-## Citation
+### Software
 
-> No formal citation. Project created for Epitech's WS2 mini-hackathon 2025.
+- PEFT 0.15.2
+- Transformers
+- Accelerate
+- Unsloth