探How 社区

HuggingFace用户xendoo发布了一个1420万参数的微型Llama模型，专门用于逐token校验vLLM等推理运行时是否与HuggingFace参考实现一致。该模型在连续的token-id序列上从零训练，序列按模50000循环，推理时用贪心解码即可精确延续序列，因此结果可人工核对。模型采用Apache-2.0许可证，目前下载量为0。

以下正文同步自 Hacker News 最新，版权归原站所有，已转换为易读排版。

Like 0

Safetensors

llama

tiny

Synthetic

sorted-tokens

runtime-verification

License: apache-2.0

Model card Files Files and versions xet

Community

Copy to bucket

Consecutive-token tiny llama (14.2M params)

Causal LM trained from scratch on consecutive token-id runs [k, k+1, ...] mod 50000 (ids 0..49999; 49999 wraps to 0). Greedy decoding continues the run exactly, which makes it a hand-checkable oracle for verifying inference runtimes (vLLM, etc.) token-for-token against the HuggingFace reference.

","eos_token":"<|endoftext|>","unk_token":"<|endoftext|>","pad_token":"<|endoftext|>"}},"createdAt":"2026-11-02T21:11:53.000Z","discussionsDisabled":false,"discussionsSorting":"recently-created","downloads":0,"downloadsAllTime":0,"id":"xendoo/sorted-tiny-llama","isLikedByUser":false,"availableInferenceProviders":[],"showHuggingChatEntry":false,"inference":"","lastModified":"2026-11-02T21:12:15.000Z","likes":0,"librariesOther":[],"trackDownloads":true,"model-index":null,"private":false,"repoType":"model","gated":false,"tags":["safetensors","llama","tiny","synthetic","sorted-tokens","runtime-verification","license:apache-2.0","region:us"],"tag_objs":[{"id":"safetensors","label":"Safetensors","type":"library"},{"id":"llama","label":"llama","type":"other","clickable":true},{"id":"tiny","label":"tiny","type":"other","clickable":true},{"id":"synthetic","label":"Synthetic","type":"other","clickable":true},{"id":"sorted-tokens","label":"sorted-tokens","type":"other","clickable":true},{"id":"runtime-verification","label":"runtime-verification","type":"other","clickable":true},{"id":"license:apache-2.0","label":"apache-2.0","type":"license"},{"type":"region","label":"🇺🇸 Region: US","id":"region:us"}],"safetensors":{"parameters":{"F32":14177792},"total":14177792,"sharded":false,"totalFileSize":56713344},"hasBlockedOids":false,"region":"us","isQuantized":false},"inferenceContextData":{"billableEntities":[],"entityName2Providers":{}},"canWrite":false}">Safetensors

Model size

14.2M params

Tensor type

F32 ·

Files info

正文由探How 从来源站点 RSS 同步，内容未经改写；遇到排版缺失或需要图片、视频时请以原文为准。

开发者发布1420万参数微型Llama模型，用于逐token校验推理运行时