Blind Spots Dataset for SmolLM2-1.7B
Model Tested
model_name = "HuggingFaceTB/SmolLM2-1.7B"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
model_name,
device_map="auto",
torch_dtype="auto"
)
def test_model(prompt, max_new_tokens=100):
inputs =… See the full description on the dataset page:
https://huggingface.co/datasets/Muhammad0981/smollm2-blindspots.