Added ability to compare results vs. llama.cpp

[AlexKoff88]

Example:

rm -rf results/smollm2_N_FP16/gt.csv
mkdir -p results/smollm2_N_FP16

# References from PyTorch FP16
wwb --base-model HuggingFaceTB/SmolLM2-360M-Instruct --gt-data results/smollm2_N_FP16/gt.csv --hf --num-samples 4


#huggingface-cli download "bartowski/SmolLM2-360M-Instruct-GGUF" "SmolLM2-360M-Instruct-f16.gguf" 
wwb --target-model models/SmolLM2-360M-Instruct-f16.gguf --gt-data results/smollm2_N_FP16/gt.csv --llamacpp --output results/smollm2_N_L_FP16 --num-samples