Jiahao Zhao

icon company Postdoc at CASIA icon location Beijing icon location 🎯 Focusing on post-training of LLM　　　　　　　　　　　　　　　　　🤗 Eager to research opportunities!

Results 40 comments of Jiahao Zhao

保存对象时出错，java.lang.IllegalStateException: unread block data

我遇到的需求是小语种文本分类，使用Stanfordnlp 实现了 Itokenizer 接口。 ```java package com.hankcs.hanlp.classification.classifiers; import java.util.ArrayList; import java.util.List; import java.util.Properties; import com.hankcs.hanlp.classification.tokenizers.ITokenizer; import edu.stanford.nlp.ling.CoreAnnotations; import edu.stanford.nlp.ling.CoreLabel; import edu.stanford.nlp.pipeline.Annotation; import edu.stanford.nlp.pipeline.StanfordCoreNLP; import edu.stanford.nlp.util.CoreMap; public class WhitespaceTokenizer implements ITokenizer...

AssertionError when run Quickstart in local envrionment

According to `asrs_strategy`, `CESimilarityMetric` is default value of `sim_metric`. So I think we could omit this `asrs_sim_metric`. https://github.com/DAI-Lab/fibber/blob/ac278a06ae3204ba54d2ba2cbdd4dbe11f0bd745/fibber/paraphrase_strategies/asrs_strategy.py#L281

Collapse reference+learner hydra heads when using LoRa

When will this feature be available?

Collapse reference+learner hydra heads when using LoRa

Cool!!! Looking forward to this update. 👍

Running into a torch.distributed.elastic.rendezvous.api.RendezvousConnectionError: The connection to the C10d store has failed.

Any update?

[紧急!!!]训练 MoE 模型建议额外安装 GroupedGEMM失败

能不能提供个docker啊

[紧急!!!]训练 MoE 模型建议额外安装 GroupedGEMM失败

@CyCle1024 求docker

DeepSeek V3 Support

See an experiment config in https://github.com/pytorch/torchtitan/blob/7d5f3cc698853d2227cf5433776406d0e0345424/torchtitan/experiments/deepseek_v3/ Does Titan support V3 training now?

Is CP support with MTP now?

Hi @Victarry, when will MCore v0.13 be released?

DeepSeekV3 lora fine tune, allocate memory too big !

Have you solved the problem?