README.md

metadata

license: apache-2.0
datasets:
  - dariolopez/ms-marco-es-500k
language:
  - es
library_name: sentence-transformers
pipeline_tag: question-answering

Model Description

The trained model is a fine-tuned version of PlanTL-GOB-ES/roberta-base-bne focused on question/answer using MS-MARCO dataset translated into Spanish.

How to use

In progress

Training

• Base Model
• Config used to train
• Dataset: dariolopez/ms-marco-es (query - positive - negative)
• Loss: TripletLoss

Config

{
    "model_name": "PlanTL-GOB-ES/roberta-base-bne",
    "max_seq_length": 512,
    "epochs": 10,
    "warmup_steps": 1000,
    "batch_size": 16,
    "optimizer_params": {
        "lr": 2e-05
    },
    "loss": "tl",
    "dataset_train_size": 500000,
    "dataset_name": "dariolopez/ms-marco-es-500k",
    "seed": 42,
    "length_embedding": 768
}