Urchade Zaratiana comments

Results 40 comments of


                                            Urchade Zaratiana

Gliner on CPU with multiple cores

`text[:512]`assume 512 characters not tokens you can change the maximum size by setting`model.config.max_len = 512`

Train on data without entities

You cannot train the model without any entity types. The model needs entity types to compute de matching scores. you can pre-define the list of labels under the key "label",...

Error: Giliner Training data

Hi, can retry without compiling the model ?

Unable to load model in offline mode

> [@baughmann](https://github.com/baughmann) The [workaround](https://github.com/urchade/GLiNER/issues/108#issuecomment-2254353271) in [#108](https://github.com/urchade/GLiNER/issues/108) should still work. Download both gliner and deberta and edit the gliner config to point to your local deberta model. I do not think...

cannot import name 'SAFETENSORS_SINGLE_FILE' from 'huggingface_hub.constants'

Hi, it looks like the problem come form your huggingface_hub version. try updating it

Question about max_types

Hi, was unavailable due to conference and seminars Yes, the model should work even >25

support chinese？

best way to know is to try it, but I think so. I have evaluate on chinese the paper but I think the token splitting (whitespace) may be suboptimal

The checkpoint you are trying to load has model type gliner but Transformers does not recognize this architecture.

```python from gliner import GLiNER # Initialize GLiNER with the base model model = GLiNER.from_pretrained("urchade/gliner_mediumv2.1") # Sample text for entity prediction text = """ Cristiano Ronaldo dos Santos Aveiro (Portuguese...

Order of labels impacts detection confidence scores

Hi, it is because the model uses positional encoding Bi-encode version of gliner of @Ingvarstep should not have this effect

flashdeberta breaks `batch_predict_entities` method

@Ingvarstep