Timeline support markdown

[Weaxs]

Is your feature request related to a problem? Please describe.
Timeline shortcode not support markdown now

Describe the solution you'd like

Describe alternatives you've considered

Additional context

[wermos]

Can you share a MWE (minimal working example)?

It might have something to do with a PR I made (#1515).

But it still seems to work normally on the Blowfish docs page.

Can you share what version of Blowfish and Hugo you were using?

[Weaxs]

Can you share a MWE (minimal working example)?

It might have something to do with a PR I made (#1515).

But it still seems to work normally on the Blowfish docs page.

Can you share what version of Blowfish and Hugo you were using?

like this:

{{< timeline >}}

{{< timelineItem icon="star" header="2023.07" badge="Jina Embeddings" >}}
[[2023.07] Jina Embeddings: A Novel Set of High-Performance Sentence Embedding Models](https://arxiv.org/abs/2307.11224):<br/> The first version of Jina's Embedding is based on the **T5 model ** ([*Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer. [2019.10]*](https://arxiv.org/abs/1910.10683)) for pre-training, and secondly, the Jina team designed a new dataset specifically for training the Embedding model.
{{< /timelineItem >}}

{{< timelineItem icon="star" header="2023.10" badge="Jina Embeddings 2" >}}
[[2023.10] Jina Embeddings 2: 8192-Token General-Purpose Text Embeddings for Long Documents](https://arxiv.org/abs/2310.19923):<br/> Jina's second version of Embedding is pre-trained on the basis of the    pretrained on the **Bert model**, using linear bias Attention ([**ALiBi**](https://arxiv.org/abs/2108.12409)) to increase the contextual information of the model input; secondly, the text data is used to fine-tune the model's embedding ability; finally, **adversarial example fine-tuning** is added to further improve the model's ability.
{{< /timelineItem >}}

{{< timelineItem icon="star" header="2024.09" badge="jina-embeddings-v3" >}}
[[2024.09] jina-embeddings-v3: Multilingual Embeddings With Task LoRA](https://arxiv.org/abs/2409.10173):<br/> Jina's third version of Embedding is a relatively new model. It is trained and fine-tuned based on the **XLM-RoBERTa model** (here, BGE-M3 is borrowed), and FlashAttention 2 is introduced to improve computational efficiency. <br/> One of the more innovative points is that the Jina team has introduced five **task-specific LoRA adapters**. These five adapters are trained independently. They are: ① **retrieval.passage**: used to embed documents in the query document retrieval task, mainly paragraph embedding in asymmetric retrieval ② **retrieval.query**: in the query query in query-document retrieval tasks, mainly query embedding in asymmetric retrieval ③ **separation**: cluster analysis of documents for clustering and reordering tasks ④ **classification**: text classification for classification tasks ⑤ **text-matching**: semantic text similarity matching for tasks designed for semantic similarity, such as symmetric retrieval.
{{< /timelineItem >}}

{{< /timeline >}}

preview at v2.78.0

[wermos]

Can you share which version number it used to work at? I can confirm that it is appearing broken for me as well, as you've shown.

[Weaxs]

Can you share which version number it used to work at? I can confirm that it is appearing broken for me as well, as you've shown.

I'm not sure which version it used to work at.

maybe it has never worked correctly ? 😶

[wermos]

maybe it has never worked correctly ?

I believe that may be the case. I can think of a fix, but then we lose the ability to embed GitHub cards inside timelines. I'm not sure how fond of that fix the creator of the theme may be.