license stringlengths 2 30 | tags stringlengths 2 513 | is_nc bool 1 class | readme_section stringlengths 201 597k | hash stringlengths 32 32 |
|---|---|---|---|---|
other | ['vision', 'image-segmentation'] | false | Mask2Former Mask2Former model trained on Cityscapes semantic segmentation (tiny-sized version, Swin backbone). It was introduced in the paper [Masked-attention Mask Transformer for Universal Image Segmentation ](https://arxiv.org/abs/2112.01527) and first released in [this repository](https://github.com/facebookresearch/Mask2Former/). Disclaimer: The team releasing Mask2Former did not write a model card for this model so this model card has been written by the Hugging Face team. | 3f956b9e1d387e4884454efaef3305b1 |
other | ['vision', 'image-segmentation'] | false | load Mask2Former fine-tuned on Cityscapes semantic segmentation processor = AutoImageProcessor.from_pretrained("facebook/mask2former-swin-tiny-cityscapes-semantic") model = Mask2FormerForUniversalSegmentation.from_pretrained("facebook/mask2former-swin-tiny-cityscapes-semantic") url = "http://images.cocodataset.org/val2017/000000039769.jpg" image = Image.open(requests.get(url, stream=True).raw) inputs = processor(images=image, return_tensors="pt") with torch.no_grad(): outputs = model(**inputs) | 735ae7c11c873a87809c91d83d8f731c |
creativeml-openrail-m | ['text-to-image'] | false | clay_icon Dreambooth model trained by andrewburns with [Hugging Face Dreambooth Training Space](https://huggingface.co/spaces/multimodalart/dreambooth-training) with the v1-5 base model You run your new concept via `diffusers` [Colab Notebook for Inference](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_dreambooth_inference.ipynb). Don't forget to use the concept prompts! Sample pictures of: clay (use that on your prompt)  | b300efefa3743a02dc0baa106cf6d10d |
apache-2.0 | ['generated_from_trainer'] | false | w2v2-ami This model is a fine-tuned version of [facebook/wav2vec2-large-960h-lv60-self](https://huggingface.co/facebook/wav2vec2-large-960h-lv60-self) on the None dataset. It achieves the following results on the evaluation set: - Loss: 0.8686 - Wer: 0.2861 | c6a69c875403203fad5b0ea9d5642d93 |
apache-2.0 | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | Wer | |:-------------:|:-----:|:----:|:---------------:|:------:| | 5.6021 | 3.07 | 500 | 2.9176 | 0.9997 | | 2.5006 | 6.13 | 1000 | 1.0535 | 0.3617 | | 0.9926 | 9.2 | 1500 | 0.8614 | 0.3036 | | 0.809 | 12.27 | 2000 | 0.8676 | 0.2921 | | 0.73 | 15.34 | 2500 | 0.8190 | 0.2966 | | 0.6658 | 18.4 | 3000 | 0.8707 | 0.2900 | | 0.6295 | 21.47 | 3500 | 0.8660 | 0.2821 | | 0.6089 | 24.54 | 4000 | 0.8767 | 0.2829 | | 0.5974 | 27.61 | 4500 | 0.8686 | 0.2861 | | d9140ce8365990a00fa8249b87ac1c12 |
apache-2.0 | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | Accuracy | |:-------------:|:-----:|:----:|:---------------:|:--------:| | No log | 1.0 | 125 | 0.4842 | 0.8129 | | 22b002bbf32de1b7ce3116102aaef746 |
apache-2.0 | ['translation'] | false | opus-mt-pis-fr * source languages: pis * target languages: fr * OPUS readme: [pis-fr](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/pis-fr/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-16.zip](https://object.pouta.csc.fi/OPUS-MT-models/pis-fr/opus-2020-01-16.zip) * test set translations: [opus-2020-01-16.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/pis-fr/opus-2020-01-16.test.txt) * test set scores: [opus-2020-01-16.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/pis-fr/opus-2020-01-16.eval.txt) | 73740c83bb12083ad248718f10d22a25 |
apache-2.0 | ['image-classification', 'timm', 'normalization-free', 'efficient-channel-attention'] | false | Model description This model variant was slimmed down from the original F0 variant in the paper for improved runtime characteristics (throughput, memory use) in PyTorch, on a GPU accelerator. It utilizes [Efficient Channel Attention (ECA)](https://arxiv.org/abs/1910.03151) instead of Squeeze-Excitation. It also features SiLU activations instead of the usual GELU. Like other models in the NF family, this model contains no normalization layers (batch, group, etc). The models make use of [Weight Standardized](https://arxiv.org/abs/1903.10520) convolutions with additional scaling values in lieu of normalization layers. | 8196ab8818fa9423c2a56140f06ed7ed |
apache-2.0 | ['image-classification', 'timm', 'normalization-free', 'efficient-channel-attention'] | false | Intended uses & limitations You can use the raw model to classify images along the 1,000 ImageNet labels, but you can also change its head to fine-tune it on a downstream task (another classification task with different labels, image segmentation or object detection, to name a few). | 256fd4101cf934baf58e2c53d4f1f0a6 |
apache-2.0 | ['image-classification', 'timm', 'normalization-free', 'efficient-channel-attention'] | false | How to use You can use this model with the usual factory method in [`timm`](https://github.com/rwightman/pytorch-image-models): ```python import PIL import timm import torch model = timm.create_model("hf_hub:timm/eca_nfnet_l0") config = model.default_cfg img_size = config["test_input_size"][-1] if "test_input_size" in config else config["input_size"][-1] transform = timm.data.transforms_factory.transforms_imagenet_eval( img_size=img_size, interpolation=config["interpolation"], mean=config["mean"], std=config["std"], crop_pct=config["crop_pct"], ) img = PIL.Image.open(path_to_an_image) img = img.convert("RGB") input_tensor = transform(cat_img) input_tensor = input_tensor.unsqueeze(0) | 35452b9d2227e79afbc25c8d4b495c5a |
apache-2.0 | ['image-classification', 'timm', 'normalization-free', 'efficient-channel-attention'] | false | Limitations and bias The training images in the dataset are usually photos clearly representing one of the 1,000 labels. The model will probably not generalize well on drawings or images containing multiple objects with different labels. The training images in the dataset come mostly from the US (45.4%) and Great Britain (7.6%). As such the model or models created by fine-tuning this model will work better on images picturing scenes from these countries (see [this paper](https://arxiv.org/abs/1906.02659) for examples). More generally, [recent research](https://arxiv.org/abs/2010.15052) has shown that even models trained in an unsupervised fashion on ImageNet (i.e. without using the labels) will pick up racial and gender bias represented in the training images. | 8bac53e6e33df211def545d17afc3b1f |
apache-2.0 | ['image-classification', 'timm', 'normalization-free', 'efficient-channel-attention'] | false | Training procedure For stability during training it is highly recommended to train all NFNet variants with gradient clipping enabled. This model was trained with an Adaptive Gradient Clipping (AGC) factor of 0.015 as described in [the paper](https://arxiv.org/abs/2102.06171). Similar to the paper, a cosine learning rate decay was employed using SGD w/ nesterov. Moderate to heavy augmentation ([RandAugment](https://arxiv.org/abs/1909.13719)) and regularization (dropout, stochastic depth) is recommended for training. | 6096545d667e8829ce9b4614de3d024b |
apache-2.0 | ['image-classification', 'timm', 'normalization-free', 'efficient-channel-attention'] | false | BibTeX entry and citation info NFNet model architecture: ```bibtex @article{brock2021high, author={Andrew Brock and Soham De and Samuel L. Smith and Karen Simonyan}, title={High-Performance Large-Scale Image Recognition Without Normalization}, journal={arXiv preprint arXiv:2102.06171}, year={2021} } ``` L0 model variant & pretraining: ```bibtex @misc{rw2019timm, author = {Ross Wightman}, title = {PyTorch Image Models}, year = {2019}, publisher = {GitHub}, journal = {GitHub repository}, doi = {10.5281/zenodo.4414861}, howpublished = {\url{https://github.com/rwightman/pytorch-image-models}} } ``` | 6bb854fb2b9128f26074ed2bf9b7504b |
apache-2.0 | ['translation'] | false | opus-mt-bzs-fi * source languages: bzs * target languages: fi * OPUS readme: [bzs-fi](https://github.com/Helsinki-NLP/OPUS-MT-train/blob/master/models/bzs-fi/README.md) * dataset: opus * model: transformer-align * pre-processing: normalization + SentencePiece * download original weights: [opus-2020-01-08.zip](https://object.pouta.csc.fi/OPUS-MT-models/bzs-fi/opus-2020-01-08.zip) * test set translations: [opus-2020-01-08.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/bzs-fi/opus-2020-01-08.test.txt) * test set scores: [opus-2020-01-08.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/bzs-fi/opus-2020-01-08.eval.txt) | 66014a265f50095a38c846a3a0ce8368 |
creativeml-openrail-m | ['text-to-image', 'stable-diffusion'] | false | beyaynetu" Dreambooth model trained by Tinsae with [TheLastBen's fast-DreamBooth](https://colab.research.google.com/github/TheLastBen/fast-stable-diffusion/blob/main/fast-DreamBooth.ipynb) notebook Test the concept via A1111 Colab [fast-Colab-A1111](https://colab.research.google.com/github/TheLastBen/fast-stable-diffusion/blob/main/fast_stable_diffusion_AUTOMATIC1111.ipynb) Sample pictures of this concept: .png) .png) .png) .png) | aed5cda925dd901270a9ac9f33f33c69 |
apache-2.0 | ['generated_from_trainer'] | false | tuto-bert-finetuned-ner This model is a fine-tuned version of [bert-base-cased](https://huggingface.co/bert-base-cased) on the conll2003 dataset. It achieves the following results on the evaluation set: - Loss: 0.0827 - Precision: 0.9380 - Recall: 0.9525 - F1: 0.9452 - Accuracy: 0.9867 | 69b5ff65c4f952f2300933374af4df24 |
apache-2.0 | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy | |:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:| | 0.0218 | 1.0 | 1756 | 0.0714 | 0.9372 | 0.9524 | 0.9447 | 0.9862 | | 0.0123 | 2.0 | 3512 | 0.0761 | 0.9347 | 0.9510 | 0.9428 | 0.9859 | | 0.0063 | 3.0 | 5268 | 0.0827 | 0.9380 | 0.9525 | 0.9452 | 0.9867 | | 2b606371ef95c5953dfb680dd02734aa |
creativeml-openrail-m | [] | false | JellyCute7 (artist) Style [Hypernetwork] Hypernetwork trained on art by artist [JellyCute7](https://www.pixiv.net/en/users/1053112). [](https://www.buymeacoffee.com/stricky) | 565554d8151bc7c95d9dfd393c494f1f |
creativeml-openrail-m | [] | false | Settings ``` Model: NAI Layer structure: (1, 2, 1) Activation function: relu Layer normalization: False Use dropout: False Raw dataset size: 208 images Final dataset size: 832 images Size: 512x512 Create flipped copies: True Split oversized images: True Captions: DeepBooru Learning rate: 0.000005 -> 13000 steps Recommended: 13000 steps ``` | 7fb027fd53c2706ba918a3a708c3e073 |
apache-2.0 | ['generated_from_trainer'] | false | finetuning-sentiment-model This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the imdb dataset. It achieves the following results on the evaluation set: - Loss: 0.2309 - Accuracy: 0.9319 - F1: 0.9323 | c1e57d445362f4523e998132907ba0b3 |
apache-2.0 | ['automatic-speech-recognition', 'en'] | false | exp_w2v2r_en_vp-100k_gender_male-0_female-10_s169 Fine-tuned [facebook/wav2vec2-large-100k-voxpopuli](https://huggingface.co/facebook/wav2vec2-large-100k-voxpopuli) for speech recognition using the train split of [Common Voice 7.0 (en)](https://huggingface.co/datasets/mozilla-foundation/common_voice_7_0). When using this model, make sure that your speech input is sampled at 16kHz. This model has been fine-tuned by the [HuggingSound](https://github.com/jonatasgrosman/huggingsound) tool. | 29e1177aaafbfa00c1950ed67d024ab3 |
apache-2.0 | [] | false | Model Description LUKE (Language Understanding with Knowledge-based Embeddings) is a new pretrained contextualized representation of words and entities based on transformer. - **Developed by:** Studio Ousia - **Shared by [Optional]:** More information needed - **Model type:** EntitySpanClassification - **Language(s) (NLP):** More information needed - **License:** Apache-2.0 - **Related Models:** [Luke-large](https://huggingface.co/studio-ousia/luke-large?text=Paris+is+the+%3Cmask%3E+of+France.) - **Parent Model:** Luke - **Resources for more information:** - [GitHub Repo](https://github.com/studio-ousia/luke) - [Associated Paper](https://arxiv.org/abs/2010.01057) | 45ab38f8d74073f2ddd83d08e793c1cb |
apache-2.0 | [] | false | Bias, Risks, and Limitations Significant research has explored bias and fairness issues with language models (see, e.g., [Sheng et al. (2021)](https://aclanthology.org/2021.acl-long.330.pdf) and [Bender et al. (2021)](https://dl.acm.org/doi/pdf/10.1145/3442188.3445922)). Predictions generated by the model may include disturbing and harmful stereotypes across protected classes; identity characteristics; and sensitive, social, and occupational groups. | e0e5945a4ea4e9289a55ca5f243e3f3d |
apache-2.0 | [] | false | Metrics LUKE achieves state-of-the-art results on five popular NLP benchmarks including * **[SQuAD v1.1](https://rajpurkar.github.io/SQuAD-explorer/)** (extractive question answering), * **[CoNLL-2003](https://www.clips.uantwerpen.be/conll2003/ner/)** (named entity recognition), **[ReCoRD](https://sheng-z.github.io/ReCoRD-explorer/)** (cloze-style question answering), * **[TACRED](https://nlp.stanford.edu/projects/tacred/)** (relation classification), and * **[Open Entity](https://www.cs.utexas.edu/~eunsol/html_pages/open_entity.html)** (entity typing). | 0c08eb0f5461bf4b934f201082e08fa1 |
apache-2.0 | [] | false | Results The experimental results are provided as follows: | Task | Dataset | Metric | LUKE-large | luke-base | Previous SOTA | | ------------------------------ | ---------------------------------------------------------------------------- | ------ | ----------------- | --------- | ------------------------------------------------------------------------- | | Extractive Question Answering | [SQuAD v1.1](https://rajpurkar.github.io/SQuAD-explorer/) | EM/F1 | **90.2**/**95.4** | 86.1/92.3 | 89.9/95.1 ([Yang et al., 2019](https://arxiv.org/abs/1906.08237)) | | Named Entity Recognition | [CoNLL-2003](https://www.clips.uantwerpen.be/conll2003/ner/) | F1 | **94.3** | 93.3 | 93.5 ([Baevski et al., 2019](https://arxiv.org/abs/1903.07785)) | | Cloze-style Question Answering | [ReCoRD](https://sheng-z.github.io/ReCoRD-explorer/) | EM/F1 | **90.6**/**91.2** | - | 83.1/83.7 ([Li et al., 2019](https://www.aclweb.org/anthology/D19-6011/)) | | Relation Classification | [TACRED](https://nlp.stanford.edu/projects/tacred/) | F1 | **72.7** | - | 72.0 ([Wang et al. , 2020](https://arxiv.org/abs/2002.01808)) | | Fine-grained Entity Typing | [Open Entity](https://www.cs.utexas.edu/~eunsol/html_pages/open_entity.html) | F1 | **78.2** | - | 77.6 ([Wang et al. , 2020](https://arxiv.org/abs/2002.01808)) | Please check the [Github repository](https://github.com/studio-ousia/luke) for more details and updates. | 1fc8b46bd795b0f513ab38a9a96d1836 |
apache-2.0 | [] | false | compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). - **Hardware Type:** More information needed - **Hours used:** More information needed - **Cloud Provider:** More information needed - **Compute Region:** More information needed - **Carbon Emitted:** More information needed | 09262095bd06e63766dd4fcc7dbcab6c |
apache-2.0 | [] | false | Citation **BibTeX:** ``` @inproceedings{yamada2020luke, title={LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention}, author={Ikuya Yamada and Akari Asai and Hiroyuki Shindo and Hideaki Takeda and Yuji Matsumoto}, booktitle={EMNLP}, year={2020} } ``` | b6cc531f3534d87e65256fed04077a59 |
apache-2.0 | [] | false | How to Get Started with the Model Use the code below to get started with the model. <details> <summary> Click to expand </summary> ```python from transformers import AutoTokenizer, LukeForEntitySpanClassification tokenizer = AutoTokenizer.from_pretrained("studio-ousia/luke-large-finetuned-conll-2003") model = LukeForEntitySpanClassification.from_pretrained("studio-ousia/luke-large-finetuned-conll-2003") ``` </details> | 99f05624bb79e393d44fe9b2707738e4 |
creativeml-openrail-m | ['text-to-image', 'safetensors', 'stable-diffusion'] | false | Broken-Mirror-Style Dreambooth model trained by Kagerage with [TheLastBen's fast-DreamBooth](https://colab.research.google.com/github/TheLastBen/fast-stable-diffusion/blob/main/fast-DreamBooth.ipynb) notebook Test the concept via A1111 Colab [fast-Colab-A1111](https://colab.research.google.com/github/TheLastBen/fast-stable-diffusion/blob/main/fast_stable_diffusion_AUTOMATIC1111.ipynb) Or you can run your new concept via `diffusers` [Colab Notebook for Inference](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_dreambooth_inference.ipynb) In my experience, the best prompting style is "brkmror cracked and shattered mirror, professional photo, descriptor of location, portrait of person, bokeh", potentially with a negative prompt of "closed eyes" Sampler DPM++ SDE Karras with at least 25 steps gives slightly better faces from my testing, CFG 7 seems fine, and since this is trained on top of 2.1, the minimum resolution is 768x768. Or course though, feel free to experiment! \(NOTE!!! This 2.1 based model is FP16, so it may not render outputs properly without Xformers\) Sample pictures of this concept:     | fb2f79feee31a1c2d64ac36e7834b82f |
mit | ['generated_from_trainer'] | false | edos-2023-baseline-microsoft-deberta-v3-base-label_vector This model is a fine-tuned version of [microsoft/deberta-v3-base](https://huggingface.co/microsoft/deberta-v3-base) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 1.5524 - F1: 0.3162 | 81750ffd23e63fcc3446a3395023e1c1 |
mit | ['generated_from_trainer'] | false | Training hyperparameters The following hyperparameters were used during training: - learning_rate: 1e-05 - train_batch_size: 32 - eval_batch_size: 32 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - lr_scheduler_warmup_steps: 5 - num_epochs: 12 - mixed_precision_training: Native AMP | 80a3afaace29ea5eb8b9a0cede238d1c |
mit | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | F1 | |:-------------:|:-----:|:----:|:---------------:|:------:| | 2.1209 | 1.18 | 100 | 1.9990 | 0.0801 | | 1.7997 | 2.35 | 200 | 1.7293 | 0.1349 | | 1.5749 | 3.53 | 300 | 1.6080 | 0.2431 | | 1.3674 | 4.71 | 400 | 1.5411 | 0.2793 | | 1.2214 | 5.88 | 500 | 1.5285 | 0.2980 | | 1.0752 | 7.06 | 600 | 1.5165 | 0.3054 | | 0.9899 | 8.24 | 700 | 1.5210 | 0.3186 | | 0.8733 | 9.41 | 800 | 1.5385 | 0.3134 | | 0.8578 | 10.59 | 900 | 1.5524 | 0.3162 | | 51ece7384cb1fb398875fa9d651a998b |
creativeml-openrail-m | ['text-to-image'] | false | Franck Dreambooth model trained by duja1 with [Hugging Face Dreambooth Training Space](https://huggingface.co/spaces/multimodalart/dreambooth-training) with the v1-5 base model You run your new concept via `diffusers` [Colab Notebook for Inference](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/sd_dreambooth_inference.ipynb). Don't forget to use the concept prompts! Sample pictures of: f123ranck (use that on your prompt) | 1872766ae07da46b81536bb9a4e3e28f |
apache-2.0 | ['image-to-text', 'image-captioning'] | false | This is an image captioning model training by Zayn ```python from transformers import VisionEncoderDecoderModel, ViTFeatureExtractor, AutoTokenizer model = VisionEncoderDecoderModel.from_pretrained("Zayn/AICVTG_What_if_a_machine_could_create_captions_automatically") feature_extractor = ViTFeatureExtractor.from_pretrained("Zayn/AICVTG_What_if_a_machine_could_create_captions_automatically") tokenizer = AutoTokenizer.from_pretrained("Zayn/AICVTG_What_if_a_machine_could_create_captions_automatically") device = torch.device("cuda" if torch.cuda.is_available() else "cpu") model.to(device) max_length = 20 num_beams = 8 gen_kwargs = {"max_length": max_length, "num_beams": num_beams} def predict_step(image_paths): images = [] for image_path in image_paths: i_image = Image.open(image_path) if i_image.mode != "RGB": i_image = i_image.convert(mode="RGB") images.append(i_image) pixel_values = feature_extractor(images=images, return_tensors="pt").pixel_values pixel_values = pixel_values.to(device) output_ids = model.generate(pixel_values, **gen_kwargs) preds = tokenizer.batch_decode(output_ids, skip_special_tokens=True) preds = [pred.strip() for pred in preds] return preds predict_step(['Image URL.jpg']) | ac81624ec7e1bcbac8eb7244033b5f27 |
apache-2.0 | ['generated_from_trainer'] | false | albert-base-v2_mnli_bc This model is a fine-tuned version of [albert-base-v2](https://huggingface.co/albert-base-v2) on the GLUE MNLI dataset. It achieves the following results on the evaluation set: - Loss: 0.2952 - Accuracy: 0.9399 | 3e7124f5e9d767d883100792f28b11c1 |
apache-2.0 | ['generated_from_trainer'] | false | Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 16 - eval_batch_size: 8 - seed: 42 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 3.0 - mixed_precision_training: Native AMP | afe3245ebcee31cd0c3e2fa04165eed2 |
apache-2.0 | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | Accuracy | |:-------------:|:-----:|:-----:|:---------------:|:--------:| | 0.2159 | 1.0 | 16363 | 0.2268 | 0.9248 | | 0.1817 | 2.0 | 32726 | 0.2335 | 0.9347 | | 0.0863 | 3.0 | 49089 | 0.3014 | 0.9401 | | 12d6be64f5a31e6867b13c921a483209 |
apache-2.0 | ['generated_from_trainer'] | false | distilbert_sa_GLUE_Experiment_logit_kd_stsb_256 This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the GLUE STSB dataset. It achieves the following results on the evaluation set: - Loss: 1.1268 - Pearson: nan - Spearmanr: nan - Combined Score: nan | 84086e2de513faa4f071ee9307889ffe |
apache-2.0 | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | Pearson | Spearmanr | Combined Score | |:-------------:|:-----:|:----:|:---------------:|:-------:|:---------:|:--------------:| | 3.1622 | 1.0 | 23 | 1.7502 | -0.0248 | -0.0193 | -0.0221 | | 1.8579 | 2.0 | 46 | 1.3087 | -0.0465 | -0.0476 | -0.0470 | | 1.3508 | 3.0 | 69 | 1.1268 | nan | nan | nan | | 1.1078 | 4.0 | 92 | 1.1974 | 0.0294 | 0.0287 | 0.0290 | | 1.0747 | 5.0 | 115 | 1.1797 | 0.0509 | 0.0597 | 0.0553 | | 1.024 | 6.0 | 138 | 1.2292 | 0.0554 | 0.0782 | 0.0668 | | 0.944 | 7.0 | 161 | 1.2819 | 0.1274 | 0.1441 | 0.1358 | | 0.795 | 8.0 | 184 | 1.2143 | 0.1987 | 0.2082 | 0.2035 | | a56cc3950289afa0f6549840c014f9e3 |
apache-2.0 | ['t5', 'translation', 'seq2seq'] | false | t5-small-24L-ccmatrix-multi A [t5-small-24L-dutch-english](https://huggingface.co/yhavinga/t5-small-24L-dutch-english) model finetuned for Dutch to English and English to Dutch translation on the CCMatrix dataset. Evaluation metrics of this model are listed in the **Translation models** section below. You can use this model directly with a pipeline for text translation: ```python model_name = "yhavinga/t5-small-24L-ccmatrix-multi" from transformers import AutoTokenizer from transformers import AutoModelForSeq2SeqLM from transformers import pipeline import torch device_num = 0 if torch.cuda.is_available() else -1 device = "cpu" if device_num < 0 else f"cuda:{device_num}" tokenizer = AutoTokenizer.from_pretrained(model_name) model = AutoModelForSeq2SeqLM.from_pretrained(model_name).to(device) params = {"max_length": 128, "num_beams": 4, "early_stopping": True} en_to_nl = pipeline("translation_en_to_nl", tokenizer=tokenizer, model=model, device=device_num) print(en_to_nl("""Young Wehling was hunched in his chair, his head in his hand. He was so rumpled, so still and colorless as to be virtually invisible.""", **params)[0]['translation_text']) nl_to_en = pipeline("translation_nl_to_en", tokenizer=tokenizer, model=model, device=device_num) print(nl_to_en("""De jonge Wehling zat gebogen in zijn stoel, zijn hoofd in zijn hand. Hij was zo stoffig, zo stil en kleurloos dat hij vrijwel onzichtbaar was.""", **params)[0]['translation_text']) ``` This **t5 eff** model has **249M** parameters. It was pre-trained with masked language modeling (denoise token span corruption) objective on the dataset `mc4_nl_cleaned` config `large_en_nl` for **1** epoch(s) and a duration of **4d10h**, with a sequence length of **512**, batch size **128** and **851852** total steps (**56B** tokens). Pre-training evaluation loss and accuracy are **1,18** and **0,74**. Refer to the evaluation section below for a comparison of the pre-trained models on summarization and translation. | 5e4a482ec03ce6129bae2434641d3598 |
apache-2.0 | ['generated_from_trainer'] | false | t5-base-extraction-cnndm_fs0.01-all This model is a fine-tuned version of [t5-base](https://huggingface.co/t5-base) on the None dataset. It achieves the following results on the evaluation set: - Loss: 1.8747 | a61179ff4109ae63723c22f56b48ac64 |
apache-2.0 | ['generated_from_trainer'] | false | Training hyperparameters The following hyperparameters were used during training: - learning_rate: 2e-05 - train_batch_size: 32 - eval_batch_size: 32 - seed: 1799 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 20 - mixed_precision_training: Native AMP | bc090ff2a722f3ca30c1e55092946a45 |
apache-2.0 | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | |:-------------:|:-----:|:----:|:---------------:| | 2.3573 | 2.25 | 200 | 1.9379 | | 1.9645 | 4.49 | 400 | 1.9068 | | 1.862 | 6.74 | 600 | 1.8823 | | 1.7958 | 8.99 | 800 | 1.8796 | | 1.7493 | 11.24 | 1000 | 1.8759 | | 1.7053 | 13.48 | 1200 | 1.8747 | | 1.6773 | 15.73 | 1400 | 1.8786 | | 1.6631 | 17.98 | 1600 | 1.8796 | | b15a1510026449077cf544d819462b93 |
apache-2.0 | ['automatic-speech-recognition', 'id'] | false | exp_w2v2t_id_hubert_s246 Fine-tuned [facebook/hubert-large-ll60k](https://huggingface.co/facebook/hubert-large-ll60k) for speech recognition using the train split of [Common Voice 7.0 (id)](https://huggingface.co/datasets/mozilla-foundation/common_voice_7_0). When using this model, make sure that your speech input is sampled at 16kHz. This model has been fine-tuned by the [HuggingSound](https://github.com/jonatasgrosman/huggingsound) tool. | fdca3583cbe034a4be26c8c26a619ff4 |
apache-2.0 | ['generated_from_trainer'] | false | distilbert_sa_GLUE_Experiment_logit_kd_mnli_96 This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the GLUE MNLI dataset. It achieves the following results on the evaluation set: - Loss: 0.5438 - Accuracy: 0.5431 | 9ecc2ddb027d94b1d66fd3c02aaccd41 |
apache-2.0 | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | Accuracy | |:-------------:|:-----:|:-----:|:---------------:|:--------:| | 0.6023 | 1.0 | 1534 | 0.5718 | 0.4960 | | 0.5673 | 2.0 | 3068 | 0.5547 | 0.5184 | | 0.5555 | 3.0 | 4602 | 0.5505 | 0.5278 | | 0.5481 | 4.0 | 6136 | 0.5466 | 0.5381 | | 0.5426 | 5.0 | 7670 | 0.5454 | 0.5403 | | 0.5382 | 6.0 | 9204 | 0.5454 | 0.5354 | | 0.5341 | 7.0 | 10738 | 0.5452 | 0.5344 | | 0.5308 | 8.0 | 12272 | 0.5428 | 0.5410 | | 0.5271 | 9.0 | 13806 | 0.5460 | 0.5451 | | 0.5239 | 10.0 | 15340 | 0.5450 | 0.5462 | | 0.5209 | 11.0 | 16874 | 0.5447 | 0.5449 | | 0.5179 | 12.0 | 18408 | 0.5452 | 0.5475 | | 0.5152 | 13.0 | 19942 | 0.5495 | 0.5454 | | 4ac928aa8a7568588ded5e3ec7343ade |
apache-2.0 | ['automatic-speech-recognition', 'zh-CN'] | false | exp_w2v2t_zh-cn_no-pretraining_s730 Fine-tuned randomly initialized wav2vec2 model for speech recognition using the train split of [Common Voice 7.0 (zh-CN)](https://huggingface.co/datasets/mozilla-foundation/common_voice_7_0). When using this model, make sure that your speech input is sampled at 16kHz. This model has been fine-tuned by the [HuggingSound](https://github.com/jonatasgrosman/huggingsound) tool. | 0d5192809b58b3e76b201baa82f7a532 |
mit | ['generated_from_trainer'] | false | gpt2-ft-with-non-challenging This model is a fine-tuned version of [gpt2](https://huggingface.co/gpt2) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 3.9906 | f10337df75d03f3a77a948970c196e21 |
mit | ['generated_from_trainer'] | false | Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.0005 - train_batch_size: 2 - eval_batch_size: 2 - seed: 42 - gradient_accumulation_steps: 32 - total_train_batch_size: 64 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - lr_scheduler_warmup_steps: 100.0 - num_epochs: 4 | 8a84757a96116ac4b1c3ef443cce7d05 |
mit | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | |:-------------:|:-----:|:----:|:---------------:| | No log | 1.0 | 1 | 4.0984 | | No log | 2.0 | 2 | 4.0802 | | No log | 3.0 | 3 | 4.0443 | | No log | 4.0 | 4 | 3.9906 | | 8f11f6e499f8384327212daa627afe5a |
apache-2.0 | ['setfit', 'sentence-transformers', 'text-classification'] | false | fathyshalab/massive_calendar-roberta-large-v1-3-93 This is a [SetFit model](https://github.com/huggingface/setfit) that can be used for text classification. The model has been trained using an efficient few-shot learning technique that involves: 1. Fine-tuning a [Sentence Transformer](https://www.sbert.net) with contrastive learning. 2. Training a classification head with features from the fine-tuned Sentence Transformer. | 69fd715273342d27a047be442f4a2e9c |
apache-2.0 | ['generated_from_trainer'] | false | finetuning-sentiment-model-3000-samples This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the imdb dataset. It achieves the following results on the evaluation set: - Loss: 0.3163 - Accuracy: 0.8667 - F1: 0.8693 | 60c2bf8310a8f1f68a5fd838b1c19b68 |
apache-2.0 | ['generated_from_trainer'] | false | distilbert-base-uncased__sst2__train-16-4 This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the None dataset. It achieves the following results on the evaluation set: - Loss: 1.1501 - Accuracy: 0.6387 | deb1e2473864c98c90cb69060e9b8abe |
apache-2.0 | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | Accuracy | |:-------------:|:-----:|:----:|:---------------:|:--------:| | 0.7043 | 1.0 | 7 | 0.7139 | 0.2857 | | 0.68 | 2.0 | 14 | 0.7398 | 0.2857 | | 0.641 | 3.0 | 21 | 0.7723 | 0.2857 | | 0.5424 | 4.0 | 28 | 0.8391 | 0.2857 | | 0.5988 | 5.0 | 35 | 0.7761 | 0.2857 | | 0.3698 | 6.0 | 42 | 0.7707 | 0.4286 | | 0.3204 | 7.0 | 49 | 0.8290 | 0.4286 | | 0.2882 | 8.0 | 56 | 0.6551 | 0.5714 | | 0.1512 | 9.0 | 63 | 0.5652 | 0.5714 | | 0.1302 | 10.0 | 70 | 0.5278 | 0.5714 | | 0.1043 | 11.0 | 77 | 0.4987 | 0.7143 | | 0.0272 | 12.0 | 84 | 0.5278 | 0.5714 | | 0.0201 | 13.0 | 91 | 0.5307 | 0.5714 | | 0.0129 | 14.0 | 98 | 0.5382 | 0.5714 | | 0.0117 | 15.0 | 105 | 0.5227 | 0.5714 | | 0.0094 | 16.0 | 112 | 0.5066 | 0.7143 | | 0.0104 | 17.0 | 119 | 0.4869 | 0.7143 | | 0.0069 | 18.0 | 126 | 0.4786 | 0.7143 | | 0.0062 | 19.0 | 133 | 0.4707 | 0.7143 | | 0.0065 | 20.0 | 140 | 0.4669 | 0.7143 | | 0.0051 | 21.0 | 147 | 0.4686 | 0.7143 | | 0.0049 | 22.0 | 154 | 0.4784 | 0.7143 | | 0.0046 | 23.0 | 161 | 0.4839 | 0.7143 | | 0.0039 | 24.0 | 168 | 0.4823 | 0.7143 | | 0.0044 | 25.0 | 175 | 0.4791 | 0.7143 | | 0.0037 | 26.0 | 182 | 0.4778 | 0.7143 | | 0.0038 | 27.0 | 189 | 0.4770 | 0.7143 | | 0.0036 | 28.0 | 196 | 0.4750 | 0.7143 | | 0.0031 | 29.0 | 203 | 0.4766 | 0.7143 | | 0.0031 | 30.0 | 210 | 0.4754 | 0.7143 | | fa2d22bb210fcfaa2464757d4985da69 |
apache-2.0 | ['generated_from_trainer'] | false | presentation_emotion_1234567 This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the tweet_eval dataset. It achieves the following results on the evaluation set: - Loss: 1.0237 - F1: 0.7273 | 6a10f39d3a71edd9d45dc937f99c7a3d |
apache-2.0 | ['generated_from_trainer'] | false | Training hyperparameters The following hyperparameters were used during training: - learning_rate: 5.18796906442746e-05 - train_batch_size: 8 - eval_batch_size: 8 - seed: 1234567 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - num_epochs: 4 | 7b60bff8dad9e13eb641a7ba88f0bcdf |
apache-2.0 | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | F1 | |:-------------:|:-----:|:----:|:---------------:|:------:| | 0.1189 | 1.0 | 408 | 0.6827 | 0.7164 | | 1.0678 | 2.0 | 816 | 0.6916 | 0.7396 | | 0.6582 | 3.0 | 1224 | 0.9281 | 0.7276 | | 0.0024 | 4.0 | 1632 | 1.0237 | 0.7273 | | c01d5f539cd6c0ad1e989b54b2e3198d |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | list-of-languages) for all available. **Purpose:** This module takes a search query [1] and a passage [2] and calculates if the passage matches the query. It can be used as an improvement for Elasticsearch Results and boosts the relevancy by up to 100%. **Architecture:** On top of BERT there is a Densly Connected NN which takes the 768 Dimensional [CLS] Token as input and provides the output ([Arxiv](https://arxiv.org/abs/1901.04085)). **Output:** Just a single value between between -10 and 10. Better matching query,passage pairs tend to have a higher a score. | 1ef1b9b71121cff05ce63c905abffe6c |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | Intended uses & limitations Both query[1] and passage[2] have to fit in 512 Tokens. As you normally want to rerank the first dozens of search results keep in mind the inference time of approximately 300 ms/query. | 49bdd42fb7a4c2dc6a99f00bc355fbd6 |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | How to use ```python from transformers import AutoTokenizer, AutoModelForSequenceClassification tokenizer = AutoTokenizer.from_pretrained("amberoad/bert-multilingual-passage-reranking-msmarco") model = AutoModelForSequenceClassification.from_pretrained("amberoad/bert-multilingual-passage-reranking-msmarco") ``` This Model can be used as a drop-in replacement in the [Nboost Library](https://github.com/koursaros-ai/nboost) Through this you can directly improve your Elasticsearch Results without any coding. | 1a91815f7bc22a7146b529d54a6ccf1b |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | Training data This model is trained using the [**Microsoft MS Marco Dataset**](https://microsoft.github.io/msmarco/ "Microsoft MS Marco"). This training dataset contains approximately 400M tuples of a query, relevant and non-relevant passages. All datasets used for training and evaluating are listed in this [table](https://github.com/microsoft/MSMARCO-Passage-Ranking | 351757c5faf8750038a5dcb33630d772 |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | data-information-and-formating). The used dataset for training is called *Train Triples Large*, while the evaluation was made on *Top 1000 Dev*. There are 6,900 queries in total in the development dataset, where each query is mapped to top 1,000 passage retrieved using BM25 from MS MARCO corpus. | 8e09678213ecbd1fbdf9abb829c4075d |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | Training procedure The training is performed the same way as stated in this [README](https://github.com/nyu-dl/dl4marco-bert "NYU Github"). See their excellent Paper on [Arxiv](https://arxiv.org/abs/1901.04085). We changed the BERT Model from an English only to the default BERT Multilingual uncased Model from [Google](https://huggingface.co/bert-base-multilingual-uncased). Training was done 400 000 Steps. This equaled 12 hours an a TPU V3-8. | ec025aadcb2383f96bb79b62c7c4d8ec |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | Eval results We see nearly similar performance than the English only Model in the English [Bing Queries Dataset](http://www.msmarco.org/). Although the training data is English only internal Tests on private data showed a far higher accurancy in German than all other available models. Fine-tuned Models | Dependency | Eval Set | Search Boost<a href=' | 390943c0d7266dd21d406381c3ad485b |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | benchmarks'> | Speed on GPU ----------------------------------------------------------------------------------- | ---------------------------------------------------------------------------- | ------------------------------------------------------------------ | ----------------------------------------------------- | ---------------------------------- **`amberoad/Multilingual-uncased-MSMARCO`** (This Model) | <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-blue"/> | <a href ='http://www.msmarco.org/'>bing queries</a> | **+61%** <sub><sup>(0.29 vs 0.18)</sup></sub> | ~300 ms/query <a href=' | 1b56d88ac43e4e208c56145965da0f71 |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | footnotes'> `nboost/pt-tinybert-msmarco` | <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-red"/> | <a href ='http://www.msmarco.org/'>bing queries</a> | **+45%** <sub><sup>(0.26 vs 0.18)</sup></sub> | ~50ms/query <a href=' | 0f65be3a6118e522daf9fbac90c35b09 |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | footnotes'> `nboost/pt-bert-base-uncased-msmarco` | <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-red"/> | <a href ='http://www.msmarco.org/'>bing queries</a> | **+62%** <sub><sup>(0.29 vs 0.18)</sup></sub> | ~300 ms/query<a href=' | 859776893ca87f111b8ee76314d2a675 |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | footnotes'> `nboost/pt-bert-large-msmarco` | <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-red"/> | <a href ='http://www.msmarco.org/'>bing queries</a> | **+77%** <sub><sup>(0.32 vs 0.18)</sup></sub> | - `nboost/pt-biobert-base-msmarco` | <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-red"/> | <a href ='https://github.com/naver/biobert-pretrained'>biomed</a> | **+66%** <sub><sup>(0.17 vs 0.10)</sup></sub> | ~300 ms/query<a href=' | 2bd4dcc3b1fd4c4023aebd60c42e1984 |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | Contact Infos  Amberoad is a company focussing on Search and Business Intelligence. We provide you: * Advanced Internal Company Search Engines thorugh NLP * External Search Egnines: Find Competitors, Customers, Suppliers **Get in Contact now to benefit from our Expertise:** The training and evaluation was performed by [**Philipp Reissel**](https://reissel.eu/) and [**Igli Manaj**](https://github.com/iglimanaj) [ Linkedin](https://de.linkedin.com/company/amberoad) | <svg xmlns="http://www.w3.org/2000/svg" x="0px" y="0px" width="32" height="32" viewBox="0 0 172 172" style=" fill: | 776e10e9a3b82b6f508893c9b8b7a923 |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | 000000;"><g fill="none" fill-rule="nonzero" stroke="none" stroke-width="1" stroke-linecap="butt" stroke-linejoin="miter" stroke-miterlimit="10" stroke-dasharray="" stroke-dashoffset="0" font-family="none" font-weight="none" font-size="none" text-anchor="none" style="mix-blend-mode: normal"><path d="M0,172v-172h172v172z" fill="none"></path><g fill=" | e32755667002493b2e707b65f2c3a8ee |
apache-2.0 | ['msmarco', 'multilingual', 'passage reranking'] | false | e67e22"><path d="M37.625,21.5v86h96.75v-86h-5.375zM48.375,32.25h10.75v10.75h-10.75zM69.875,32.25h10.75v10.75h-10.75zM91.375,32.25h32.25v10.75h-32.25zM48.375,53.75h75.25v43h-75.25zM80.625,112.875v17.61572c-1.61558,0.93921 -2.94506,2.2687 -3.88428,3.88428h-49.86572v10.75h49.86572c1.8612,3.20153 5.28744,5.375 9.25928,5.375c3.97183,0 7.39808,-2.17347 9.25928,-5.375h49.86572v-10.75h-49.86572c-0.93921,-1.61558 -2.2687,-2.94506 -3.88428,-3.88428v-17.61572z"></path></g></g></svg>[Homepage](https://de.linkedin.com/company/amberoad) | [Email](info@amberoad.de) | 7fcefff3e778233ca10d301c4a90add4 |
mit | ['reward-model', 'reward_model', 'RLHF'] | false | Reward model trained from human feedback Reward model (RM) trained to predict which generated answer is better judged by a human, given a question. RM are useful in these domain: - QA model evaluation - serves as reward score in RLHF - detect potential toxic response via ranking All models are train on these dataset with a same split seed across datasets (if validation split wasn't available) - [webgpt_comparisons](https://huggingface.co/datasets/openai/webgpt_comparisons) - [summarize_from_feedback](https://huggingface.co/datasets/openai/summarize_from_feedback) - [synthetic-instruct-gptj-pairwise](https://huggingface.co/datasets/Dahoas/synthetic-instruct-gptj-pairwise) - [anthropic_hh-rlhf](https://huggingface.co/datasets/Anthropic/hh-rlhf) | ab2994995a065ffb329cfb5dae4f2c68 |
mit | ['reward-model', 'reward_model', 'RLHF'] | false | How to use ``` from transformers import AutoModelForSequenceClassification, AutoTokenizer reward_name = "OpenAssistant/reward-model-deberta-v3-large-v2" rank_model, tokenizer = AutoModelForSequenceClassification.from_pretrained(reward_name), AutoTokenizer.from_pretrained(reward_name) question, answer = "Explain nuclear fusion like I am five", "Nuclear fusion is the process by which two or more protons and neutrons combine to form a single nucleus. It is a very important process in the universe, as it is the source of energy for stars and galaxies. Nuclear fusion is also a key process in the production of energy for nuclear power plants." inputs = tokenizer(question, answer, return_tensors='pt') score = rank_model(**inputs).logits[0].cpu().detach() print(score) ``` **Toxic response detection** ``` from transformers import AutoModelForSequenceClassification, AutoTokenizer reward_name = "OpenAssistant/reward-model-deberta-v3-large-v2" rank_model, tokenizer = AutoModelForSequenceClassification.from_pretrained(reward_name), AutoTokenizer.from_pretrained(reward_name) question = "I just came out of from jail, any suggestion of my future?" helpful = "It's great to hear that you have been released from jail." bad = "Go back to jail you scum" inputs = tokenizer(question, helpful, return_tensors='pt') good_score = rank_model(**inputs).logits[0].cpu().detach() inputs = tokenizer(question, bad, return_tensors='pt') bad_score = rank_model(**inputs).logits[0].cpu().detach() print(good_score > bad_score) | aebe96a98648e2af397d78d4d307fe3f |
mit | ['reward-model', 'reward_model', 'RLHF'] | false | Performance Validation split accuracy | Model | [WebGPT](https://huggingface.co/datasets/openai/webgpt_comparisons) | [Summary](https://huggingface.co/datasets/openai/summarize_from_feedback) | [SytheticGPT](https://huggingface.co/datasets/Dahoas/synthetic-instruct-gptj-pairwise) | [Anthropic RLHF]() | |---|---|---|---|---| | [electra-large-discriminator](https://huggingface.co/OpenAssistant/reward-model-electra-large-discriminator) | 59.30 | 68.66 | 99.85 | 54.33 | | **[deberta-v3-large-v2](https://huggingface.co/OpenAssistant/reward-model-deberta-v3-large-v2)** | **61.57** | 71.47 | 99.88 | **69.25** | | [deberta-v3-large](https://huggingface.co/OpenAssistant/reward-model-deberta-v3-large) | 61.13 | 72.23 | **99.94** | 55.62 | | [deberta-v3-base](https://huggingface.co/OpenAssistant/reward-model-deberta-v3-base) | 59.07 | 66.84 | 99.85 | 54.51 | | deberta-v2-xxlarge | 58.67 | **73.27** | 99.77 | 66.74 | Its likely SytheticGPT has somekind of surface pattern on the choosen-rejected pair which makes it trivial to differentiate between better the answer. | d567460ea44233e0258949e270741207 |
mit | ['reward-model', 'reward_model', 'RLHF'] | false | Other Sincere thanks to [stability.ai](https://stability.ai/) for their unwavering support in terms of A100 computational resources. Their contribution was crucial in ensuring the smooth completion of this research project. | 5d18e4efdca932a0868bfb6aaa17c1a7 |
apache-2.0 | [] | false | 简介 Brief Introduction 基于simcse无监督版本，用搜集整理的中文NLI数据进行simcse有监督任务的训练。在中文句子对任务上有良好的效果。 **Erlangshen-SimCSE-110M-Chinese** is based on the unsupervised version of simcse, And training simcse supervised task with collected and sorted chinese NLI data for. It has good effect on the task in Chinese sentences pair. | 9c0a50e808043035ce58438ed6ab5fa9 |
apache-2.0 | [] | false | 模型分类 Model Taxonomy | 需求 Demand | 任务 Task | 系列 Series | 模型 Model | 参数 Parameter | 额外 Extra | | :----: | :----: | :----: | :----: | :----: | :----: | | 通用 General | 自然语言生成 NLU | 二郎神 Erlangshen | Bert | 110M | 中文 Chinese | | 93bc6c59d6131ea050bb857a6832aa47 |
apache-2.0 | [] | false | 模型信息 Model Information 为了获得一个通用句子向量表征的模型，我们基于bert-base模型用了大量的无监督数据和有监督数据进行对比学习，最终获得了一个无需微调就能够利用模型输出的[CLS]进行相似度判断的模型。与用bert模型在针对任务微调后，再进行句子相似度任务不同，我们的模型在预训练完成后直接具备提取句子向量的能力。在一些任务上有如下的测评效果： In order to obtain a general sentence-embedding-model, we use a large number of unsupervised data and supervised data for comparative learning based on the Bert-base model, and finally obtained a model that can use the [CLS] output from the model to judge the similarity without fine-tuning. Different from the sentence similarity task after fine tuning the task with the bert model, our model has the ability to extract sentence vectors directly after pre training. In some tasks, the evaluation results are as follows: |模型 | LCQMC | BQ | PAWSX | ATEC | STS-B | | :----: | :----: | :----: | :----: | :----: | :----: | |Bert | 62 |38.62 |17.38 |28.98 |68.27| |Bert-large| 63.78| 37.51| 18.63| 30.24| 68.87| |RoBerta| 67.3| 39.89| 16.79| 30.57| 69.|36 |RoBerta large |67.25 |38.39 |19.09 |30.85 |69.36| |RoFormer| 63.58| 39.9 |17.52| 29.37 |67.32| |SimBERT| 73.43| 40.98| 15.87| 31.24| 72| |Erlangshen-SimCSE-110M-Chinese|74.94| 56.97| 21.84| 34.12| 70.5| *备注:我们的模型是直接用[cls]，无whitening；其余模型是last avg + whitening* *ps:Our model use [cls] directly，and no whitening；Other model use last avg and do whitening* | 19f14baf6ff6b1097f0558e9a2d26daf |
apache-2.0 | [] | false | 加载模型 Loading Models ```python from transformers import AutoTokenizer,AutoModelForMaskedLM model =AutoModelForMaskedLM.from_pretrained('IDEA-CCNL/Erlangshen-SimCSE-110M-Chinese') tokenizer = AutoTokenizer.from_pretrained('IDEA-CCNL/Erlangshen-SimCSE-110M-Chinese') ``` | 59d7146b30cf6ae62e1da21a555438d5 |
apache-2.0 | [] | false | 使用示例 Usage Examples ```python import torch from sklearn.metrics.pairwise import cosine_similarity texta = '今天天气真不错，我们去散步吧！' textb = '今天天气真糟糕，还是在宅家里写bug吧！' inputs_a = tokenizer(texta,return_tensors="pt") inputs_b = tokenizer(textb,return_tensors="pt") outputs_a = model(**inputs_a ,output_hidden_states=True) texta_embedding = outputs_a.hidden_states[-1][:,0,:].squeeze() outputs_b = model(**inputs_b ,output_hidden_states=True) textb_embedding = outputs_b.hidden_states[-1][:,0,:].squeeze() | afe0a01bc7782ba85b528c6898daaa3b |
apache-2.0 | ['automatic-speech-recognition', 'en'] | false | exp_w2v2t_en_vp-nl_s980 Fine-tuned [facebook/wav2vec2-large-nl-voxpopuli](https://huggingface.co/facebook/wav2vec2-large-nl-voxpopuli) for speech recognition on English using the train split of [Common Voice 7.0](https://huggingface.co/datasets/mozilla-foundation/common_voice_7_0). When using this model, make sure that your speech input is sampled at 16kHz. This model has been fine-tuned by the [HuggingSound](https://github.com/jonatasgrosman/huggingsound) tool. | fffe392e15d15f79d3754cfdecb3509d |
apache-2.0 | ['generated_from_trainer'] | false | wav2vec2-hindi-3 This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the None dataset. It achieves the following results on the evaluation set: - Loss: 2.0900 - Wer: 0.7281 | 96bcbc62d239bc5c321cf9208943a5da |
apache-2.0 | ['generated_from_trainer'] | false | Training results | Training Loss | Epoch | Step | Validation Loss | Wer | |:-------------:|:-----:|:----:|:---------------:|:------:| | 4.609 | 6.41 | 1000 | 1.2290 | 0.7497 | | 0.3754 | 12.82 | 2000 | 1.5350 | 0.7128 | | 0.1587 | 19.23 | 3000 | 1.8671 | 0.7322 | | 0.103 | 25.64 | 4000 | 1.9383 | 0.7300 | | 0.0761 | 32.05 | 5000 | 2.0767 | 0.7306 | | 0.0616 | 38.46 | 6000 | 2.0900 | 0.7281 | | 7225ac1bdbe77f45d58aa91aebfa29a8 |
apache-2.0 | ['translation'] | false | nld-epo * source group: Dutch * target group: Esperanto * OPUS readme: [nld-epo](https://github.com/Helsinki-NLP/Tatoeba-Challenge/tree/master/models/nld-epo/README.md) * model: transformer-align * source language(s): nld * target language(s): epo * model: transformer-align * pre-processing: normalization + SentencePiece (spm4k,spm4k) * download original weights: [opus-2020-06-16.zip](https://object.pouta.csc.fi/Tatoeba-MT-models/nld-epo/opus-2020-06-16.zip) * test set translations: [opus-2020-06-16.test.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/nld-epo/opus-2020-06-16.test.txt) * test set scores: [opus-2020-06-16.eval.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/nld-epo/opus-2020-06-16.eval.txt) | 1a063949bd10d65c89dcc8df996bd2a0 |
apache-2.0 | ['translation'] | false | System Info: - hf_name: nld-epo - source_languages: nld - target_languages: epo - opus_readme_url: https://github.com/Helsinki-NLP/Tatoeba-Challenge/tree/master/models/nld-epo/README.md - original_repo: Tatoeba-Challenge - tags: ['translation'] - languages: ['nl', 'eo'] - src_constituents: {'nld'} - tgt_constituents: {'epo'} - src_multilingual: False - tgt_multilingual: False - prepro: normalization + SentencePiece (spm4k,spm4k) - url_model: https://object.pouta.csc.fi/Tatoeba-MT-models/nld-epo/opus-2020-06-16.zip - url_test_set: https://object.pouta.csc.fi/Tatoeba-MT-models/nld-epo/opus-2020-06-16.test.txt - src_alpha3: nld - tgt_alpha3: epo - short_pair: nl-eo - chrF2_score: 0.355 - bleu: 16.1 - brevity_penalty: 0.9359999999999999 - ref_len: 72293.0 - src_name: Dutch - tgt_name: Esperanto - train_date: 2020-06-16 - src_alpha2: nl - tgt_alpha2: eo - prefer_old: False - long_pair: nld-epo - helsinki_git_sha: 480fcbe0ee1bf4774bcbe6226ad9f58e63f6c535 - transformers_git_sha: 2207e5d8cb224e954a7cba69fa4ac2309e9ff30b - port_machine: brutasse - port_time: 2020-08-21-14:41 | a143329b152303f3abc5f19858daf718 |
apache-2.0 | ['audio', 'automatic-speech-recognition', 'speech', 'xlsr-fine-tuning-week'] | false | Wav2Vec2-Large-XLSR-53-lg Fine-tuned [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on Luganda using the [Common Voice](https://huggingface.co/datasets/common_voice) dataset, using train, validation and other (excluding voices that are in the test set), and taking the test data for validation as well as test. When using this model, make sure that your speech input is sampled at 16kHz. | cd13fc40dcc979f2881b6591a3178d5e |
apache-2.0 | ['audio', 'automatic-speech-recognition', 'speech', 'xlsr-fine-tuning-week'] | false | Usage The model can be used directly (without a language model) as follows: ```python import torch import torchaudio from datasets import load_dataset from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor test_dataset = load_dataset("common_voice", "lg", split="test[:2%]") processor = Wav2Vec2Processor.from_pretrained("lucio/wav2vec2-large-xlsr-luganda") model = Wav2Vec2ForCTC.from_pretrained("lucio/wav2vec2-large-xlsr-luganda") resampler = torchaudio.transforms.Resample(48_000, 16_000) | 2f8679667fd6975bf7a49a655f1af755 |
apache-2.0 | ['audio', 'automatic-speech-recognition', 'speech', 'xlsr-fine-tuning-week'] | false | We need to read the audio files as arrays def speech_file_to_array_fn(batch): speech_array, sampling_rate = torchaudio.load(batch["path"]) batch["speech"] = resampler(speech_array).squeeze().numpy() return batch test_dataset = test_dataset.map(speech_file_to_array_fn) inputs = processor(test_dataset[:2]["speech"], sampling_rate=16_000, return_tensors="pt", padding=True) with torch.no_grad(): logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits predicted_ids = torch.argmax(logits, dim=-1) print("Prediction:", processor.batch_decode(predicted_ids)) print("Reference:", test_dataset["sentence"][:2]) ``` | eefc0953f5acd0c745a6d292e4030e14 |
apache-2.0 | ['audio', 'automatic-speech-recognition', 'speech', 'xlsr-fine-tuning-week'] | false | Evaluation The model can be evaluated as follows on the Luganda test data of Common Voice. (Available in Colab [here](https://colab.research.google.com/drive/1XxZ3mJOEXwIn-QH3C23jD_Qpom9aA1vH?usp=sharing).) ```python import torch import torchaudio from datasets import load_dataset, load_metric from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor import re import unidecode test_dataset = load_dataset("common_voice", "lg", split="test") wer = load_metric("wer") processor = Wav2Vec2Processor.from_pretrained("lucio/wav2vec2-large-xlsr-luganda") model = Wav2Vec2ForCTC.from_pretrained("lucio/wav2vec2-large-xlsr-luganda") model.to("cuda") chars_to_ignore_regex = '[\[\],?.!;:%"“”(){}‟ˮʺ″«»/…‽�–]' resampler = torchaudio.transforms.Resample(48_000, 16_000) | 1e18b60c720033756413a8714682c341 |
apache-2.0 | ['audio', 'automatic-speech-recognition', 'speech', 'xlsr-fine-tuning-week'] | false | We need to read the audio files as arrays def speech_file_to_array_fn(batch): speech_array, sampling_rate = torchaudio.load(batch["path"]) batch["speech"] = resampler(speech_array).squeeze().numpy() return batch def remove_special_characters(batch): | b2c38171bcb200d0c9e236935297426e |
apache-2.0 | ['audio', 'automatic-speech-recognition', 'speech', 'xlsr-fine-tuning-week'] | false | most other punctuation is ignored batch["norm_text"] = re.sub(chars_to_ignore_regex, "", batch["norm_text"]).lower().strip() batch["norm_text"] = re.sub(r"(-|' | '| +)", " ", batch["norm_text"]) | aea2d263065fb64825d5b287d8fa55ea |
apache-2.0 | ['audio', 'automatic-speech-recognition', 'speech', 'xlsr-fine-tuning-week'] | false | remove accents from a few characters (from loanwords, not tones) batch["norm_text"] = unidecode.unidecode(batch["norm_text"]) return batch test_dataset = test_dataset.map(speech_file_to_array_fn) test_dataset = test_dataset.map(remove_special_characters) def evaluate(batch): inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True) with torch.no_grad(): logits = model(inputs.input_values.to("cuda"), attention_mask=inputs.attention_mask.to("cuda")).logits pred_ids = torch.argmax(logits, dim=-1) batch["pred_strings"] = processor.batch_decode(pred_ids) return batch result = test_dataset.map(evaluate, batched=True, batch_size=8) print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["norm_text"]))) ``` **Test Result**: 29.52 % | ce4a4fd0faff2d33f95dee2cef4eba3e |
apache-2.0 | ['audio', 'automatic-speech-recognition', 'speech', 'xlsr-fine-tuning-week'] | false | Training The Common Voice `train`, `validation` and `other` datasets were used for training, excluding voices that are in both the `other` and `test` datasets. The data was augmented to twice the original size with added noise and manipulated pitch, phase and intensity. Training proceeded for 60 epochs, on 1 V100 GPU provided by OVHcloud. The `test` data was used for validation. The [script used for training](https://github.com/serapio/transformers/blob/feature/xlsr-finetune/examples/research_projects/wav2vec2/run_common_voice.py) is adapted from the [example script provided in the transformers repo](https://github.com/huggingface/transformers/blob/master/examples/research_projects/wav2vec2/run_common_voice.py). | b7a647fca73555533ebd9da02f830aec |
apache-2.0 | ['image-classification', 'timm'] | false | Model Details - **Model Type:** Image classification / feature backbone - **Model Stats:** - Params (M): 50.2 - GMACs: 8.7 - Activations (M): 21.6 - Image size: 224 x 224 - **Papers:** - A ConvNet for the 2020s: https://arxiv.org/abs/2201.03545 - **Original:** https://github.com/facebookresearch/ConvNeXt - **Dataset:** ImageNet-1k | 910c17c27d386a10d322a231e21566fe |
apache-2.0 | ['image-classification', 'timm'] | false | Image Classification ```python from urllib.request import urlopen from PIL import Image import timm img = Image.open( urlopen('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png')) model = timm.create_model('convnext_small.fb_in1k', pretrained=True) model = model.eval() | a350b5191c01ea021472a819b2a5d4e8 |
apache-2.0 | ['image-classification', 'timm'] | false | Feature Map Extraction ```python from urllib.request import urlopen from PIL import Image import timm img = Image.open( urlopen('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png')) model = timm.create_model( 'convnext_small.fb_in1k', pretrained=True, features_only=True, ) model = model.eval() | 4e1891317fea6dc70f6b8b9efbaa1178 |
apache-2.0 | ['image-classification', 'timm'] | false | Image Embeddings ```python from urllib.request import urlopen from PIL import Image import timm img = Image.open( urlopen('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png')) model = timm.create_model( 'convnext_small.fb_in1k', pretrained=True, num_classes=0, | 502fb5931ecc6c59139ae369fff31f64 |
mit | ['generated_from_keras_callback', 'text_generator'] | false | italian-literature-model-mini This model is a fine-tuned version of [gpt2](https://huggingface.co/gpt2) on an unknown dataset. It achieves the following results on the evaluation set: - Train Loss: 5.7067 - Validation Loss: 5.6842 - Epoch: 2 | ae2ec2a5f1b1346dbff958075a7759e9 |
mit | ['generated_from_keras_callback', 'text_generator'] | false | Training hyperparameters The following hyperparameters were used during training: - optimizer: {'name': 'AdamWeightDecay', 'learning_rate': {'class_name': 'WarmUp', 'config': {'initial_learning_rate': 5e-05, 'decay_schedule_fn': {'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 5e-05, 'decay_steps': 15686, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}, '__passive_serialization__': True}, 'warmup_steps': 1000, 'power': 1.0, 'name': None}}, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False, 'weight_decay_rate': 0.01} - training_precision: mixed_float16 | 3b548e8e8d8d785fe4d374e96aea9217 |
mit | ['generated_from_keras_callback', 'text_generator'] | false | Training results | Train Loss | Validation Loss | Epoch | |:----------:|:---------------:|:-----:| | 5.7065 | 5.6842 | 0 | | 5.7065 | 5.6842 | 1 | | 5.7067 | 5.6842 | 2 | | 7e69aaec066ce4eb9fbdc8977fbea0ee |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.