Update README.md

README.md

@@ -1,21 +1,31 @@
 ---
 library_name: peft
 ---
-## Training procedure
+### Training details 
 
+- Prompt tokenisation: [LlamaTokenizer](https://huggingface.co/docs/transformers/model_doc/llama2#transformers.LlamaTokenizer). 
+- The maximum context length is limited to 1,204.
+- Per device train batch: 1 
+- Gradient accumulation: 128 steps (achieving the equivalent batch_size of 128)
+- Quantisation: 8-bit (
+- Optimiser: adamw
+- Learning_rate: 3 × 10−4
+- warmup_steps: 100
+- epochs: 5
 
-The following `bitsandbytes` quantization config was used during training:
-- load_in_8bit: True
-- load_in_4bit: False
-- llm_int8_threshold: 6.0
-- llm_int8_skip_modules: None
-- llm_int8_enable_fp32_cpu_offload: False
-- llm_int8_has_fp16_weight: False
-- bnb_4bit_quant_type: fp4
-- bnb_4bit_use_double_quant: False
-- bnb_4bit_compute_dtype: float32
+- Low Rank Adaptation (LoRA)
+  - rank: 16
+  - alpha: 16
+  - dropout: 0.05
+  - target modules:  q_proj, k_proj, v_proj, and o_proj
 
+This setup reduces the trainable parameters to 26,214,400 or 0.2% of the base [Llama 2 13B Chat](https://huggingface.co/docs/transformers/model_doc/llama2) model. 
 
-### Framework versions
+### Training hardware
 
-- PEFT 0.4.0
+This model is trained on commodity hardware equipped with a: 
+- 13th Gen Intel(R) Core(TM) i7-13700KF CPU at 3.40 GHz
+- 64 GB installed RAM
+- NVIDIA GeForce RTX 4090 GPU with 24 GB onboard RAM.
+
+The trained model consumed 100 GPU hours during training.