ModelCloud/GPTQModel

gptqmodel-lazy-load-smoke-testing

End-to-end smoke test for GPT-QModel's lazy checkpoint load (LazyTurtle) and first-layer materialization through StageInputsCapture.cache_inputs. Use when verifying lazy-load behavior, first-layer input capture, shell_module_materialize module_path…

Langue du texte source : anglais

gptqmodel-tokenizer-normalization

Diagnose and correct GPT-QModel tokenizer initialization, tokenization normalization, special-token handling, prompt rendering, and chat-template problems. Use when inference or evaluation quality suggests wrong token IDs or prompts, when a model needs…

Langue du texte source : anglais