Bug in the L4 notebook (Optimizing a Model with LLM Compressor)

Hello, Cedric said in the video (9:20) that each window overlaps with the previous one, but actually there is no overlap in the code (see below) and “target_slice[:, :-trg_len] = -100” does nothing. def calculate_perplexity(model, tokenizer, dataset, max_tokens=5000, stride=512): encodings = tokeniz

Related Stories

C2 W3 Lab1 Error or mistake in code cell 2.4

📰

Measuring Latency: Key Metrics for Streaming LLM Responses

📰

Transcrevendo áudio e gerando capítulos com IA (Whisper + GPT) sem estourar o custo

Nvidia Backs Verkada's Ambitious Push into AI-Powered Physical Security