Phocinae/Phocinae-Largha-150M-v1
text-classification model by Phocinae
Older snapshot: check the original source for current information.
From the publisher
Model card & documentation
Source previewRead the publisher’s intended use, setup instructions, evaluations and limitations. The original model card is the source of truth.
斑海豹 · Phocinae-Largha-150M-v1 小海豹，大决断。 / Tiny model. Big decisions. 一斑见全豹，一点定全局。 / Spotted seal. Spot-on calls. Phocinae-Largha-150M-v1（斑海豹 Largha，jhu-clsp/mmBERT-small 微调）是一个 144.3M 参数的双语「类型化决策模型」，不是聊天模型： 输入 state 状态文本 + 一组带类型的提问（noul 是否 · choice 多选一 · score 2–10 打分），它在一次前向传播内返回带标定置信度的答案，输出零 token，且对选项顺序不敏感。 Phocinae-Largha-150M-v1 (Largha) is a 144.3M-parameter bilingual typed decision model — not a chatbot. Given a state plus typed questions (noul yes/no · choice pick-one · score 2–10), it returns calibrated answers with confidence in a single forward pass, emitting zero output tokens. 上表「同协议对照」那一行不是同类测量：我们的英文分是在该协议上拟合出来的 specialist，而 Laya / JEV / meraGPT 那几行不是。该数据集的 gold 是约 4B 级教师模型三次采样的均值，基准自己写明教师自一致性上限 0.735——这个分数衡量「与教师的一致程度」，不是「正确性」。 Our en row is a fitted specialist on this protocol; the Laya / JEV / meraGPT rows are not the same measurement class. The benchmark's stated teacher self-agreement ceiling is 0.735 — these scores measure agreement with that teacher, not correctness. 一次请求最多 64 题。注意上下文预算：编码器支持 8192 位置，但出厂 maxlen 为 512，会截断整条序列（状态 + 提问 + 选项合计）；headmaxlen 192 是决策头的注意力窗口。按 512 而不是 8192 规划你的输入。 协议全文见 docs/protocol.md。…
Read the full model card ↗ · Preview checked 2026-10-29T00:14:40.535Z
Inside the original model card — Document outline
- 一句话定义 / What is this model?
- 速览 / At a glance
- 口径警告 / Cross-class warning
- 支持的题型 / The decision protocol
- 快速开始 / Quick start
- 已知缺陷：默认渲染路径复现不出头条数字 / Known defect
- 同类位置：这一档并不空 / The 150M slot is crowded
- 为什么把决策层放在本地 / Why run it locally
- 硬件 / Hardware
- 常见问题 / FAQ
- 什么时候该用它，什么时候用大模型？
- 能省多少钱？
Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.
Links from the model card
References supplied by the publisher, not independently verified endorsements. Check the destination before downloading files or entering credentials.
Documentation belongs to its respective authors. Reported project/model license: apache-2.0. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.
Model overview
Phocinae/Phocinae-Largha-150M-v1 is a text-classification model repository published by Phocinae on Hugging Face. The source reports the transformers library.
This page summarizes Hub metadata. For intended use, training data, evaluation results and limitations, consult the original model card.
Model facts
- Task
- text-classification
- Library
- transformers
- Recent downloads (30 days)
- 0
- Cumulative likes
- 77
- Hugging Face trending score
- 68
- Reported safetensors parameters
- 144,292,870
- Architecture
- ModernBertForMaskedLM
- License
- apache-2.0
- Access
- Not gated by Hugging Face
- Created
- 2026-10-08T00:30:10.000Z
- Last modified
- 2026-10-09T05:04:34.000Z
Compatibility and lineage
Reported languages: en, zh
Reported base models: jhu-clsp/mmBERT-small
Parameter count is not a RAM/VRAM requirement. Check precision, quantization, context length and runtime compatibility in the model card; no hardware or API-cost claim is inferred here.
Use and evaluate
Check license terms and access requirements first. Review the model card’s documented loading instructions, evaluations and safety limitations. Benchmark scores are not imported by this directory, and popularity is not an accuracy ranking.
No model weights are downloaded or executed by AltAPIs. Never enable remote model code without reviewing it.
Source and freshness
Source: Hugging Face Hub. Metadata observed 2026-10-09T06:39:48.873Z. Daily imports are snapshots, not real-time monitoring.
Popularity and source listings do not establish security, suitability, licensing rights or benchmark performance.