openai

Disclaimer: The model card is taken and modified from the official CLIP repository, it can be found here. The CLIP model was developed by researchers at OpenAI to learn about what contributes to robustness in computer vision tasks. The model was also developed to test the ability of models to generalize to arbitrary image classification tasks in a zero-shot manner. It was not developed for general model deployment - to deploy models like CLIP, researchers will first need to carefully study their capabilities in relation to the specific context they’re being deployed within. The model uses a ViT-B/32 Transformer architecture as an image encoder and uses a masked self-attention Transformer as a text encoder. These encoders are trained to maximize the similarity of (image, text) pairs via a contrastive loss. The original implementation had two variants: one using a ResNet image encoder and the other using a Vision Transformer. This repository has the variant with the Vision Transformer. The model is intended as a research output for research communities. We hope that this model will enable researchers to better understand and explore zero-shot, arbitrary image classification. We also hope it can be used for interdisciplinary studies of the potential impact of such models - the CLIP paper includes a discussion of potential downstream impacts to provide an example for this sort of analysis. The primary intended users of these models are AI researchers. We primarily imagine the model will be used by researchers to better understand robustness, generalization, and other capabilities, biases, and constraints of computer vision models. Any deployed use case of the model - whether commercial or not - is currently out of scope. Non-deployed use cases such as image search in a constrained environment, are also not recommended unless there is thorough in-domain testing of the model with a specific, fixed class taxonomy. This is because our safety assessment demonstrated a high need for task specific testing especially given the variability of CLIP’s performance with different class taxonomies. This makes untested and unconstrained deployment of the model in any use case currently potentially harmful. Certain use cases which would fall under the domain of surveillance and facial recognition are always out-of-scope regardless of performance of the model. This is because the use of artificial intelligence for tasks such as these can be premature currently given the lack of testing norms and checks to ensure its fair use. Since the model has not been purposefully trained in or evaluated on any languages other than English, its use should be limited to English language use cases. The model was trained on publicly available image-caption data. This was done through a combination of crawling a handful of websites and using commonly-used pre-existing image datasets such as YFCC100M. A large portion of the data comes from our crawling of the internet. This means that the data is more representative of people and societies most connected to the internet which tend to skew towards more developed nations, and younger, male users. Our goal with building this dataset was to test out robustness and generalizability in computer vision tasks. As a result, the focus was on gathering large quantities of data from different publicly-available internet data sources. The data was gathered in a mostly non-interventionist manner. However, we only crawled websites that had policies against excessively violent and adult images and allowed us to filter out such content. We do not intend for this dataset to be used as the basis for any commercial or deployed model and will not be releasing the dataset. We have evaluated the performance of CLIP on a wide range of benchmarks across a variety of computer vision datasets such as OCR to texture recognition to fine-grained classification. The paper describes model performance on the following datasets: - Food101 - CIFAR10 - CIFAR100 - Birdsnap - SUN397 - Stanford Cars - FGVC Aircraft - VOC2007 - DTD - Oxford-IIIT Pet dataset - Caltech101 - Flowers102 - MNIST - SVHN - IIIT5K - Hateful Memes - SST-2 - UCF101 - Kinetics700 - Country211 - CLEVR Counting - KITTI Distance - STL-10 - RareAct - Flickr30 - MSCOCO - ImageNet - ImageNet-A - ImageNet-R - ImageNet Sketch - ObjectNet (ImageNet Overlap) - Youtube-BB - ImageNet-Vid CLIP and our analysis of it have a number of limitations. CLIP currently struggles with respect to certain tasks such as fine grained classification and counting objects. CLIP also poses issues with regards to fairness and bias which we discuss in the paper and briefly in the next section. Additionally, our approach to testing CLIP also has an important limitation- in many cases we have used linear probes to evaluate the performance of CLIP and there is evidence suggesting that linear probes can underestimate model performance. We find that the performance of CLIP - and the specific biases it exhibits - can depend significantly on class design and the choices one makes for categories to include and exclude. We tested the risk of certain kinds of denigration with CLIP by classifying images of people from Fairface into crime-related and non-human animal categories. We found significant disparities with respect to race and gender. Additionally, we found that these disparities could shift based on how the classes were constructed. (Details captured in the Broader Impacts Section in the paper). We also tested the performance of CLIP on gender, race and age classification using the Fairface dataset (We default to using race categories as they are constructed in the Fairface dataset.) in order to assess quality of performance across different demographics. We found accuracy >96% across all races for gender classification with ‘Middle Eastern’ having the highest accuracy (98.4%) and ‘White’ having the lowest (96.5%). Additionally, CLIP averaged ~93% for racial classification and ~63% for age classification. Our use of evaluations to test for gender, race and age classification as well as denigration harms is simply to evaluate performance of the model across people and surface potential risks and not to demonstrate an endorsement/enthusiasm for such tasks. Where to send questions or comments about the model

Disclaimer: The model card is taken and modified from the official CLIP repository, it can be found here. The CLIP model was developed by researchers at OpenAI to learn about what contributes to robustness in computer vision tasks. The model was also developed to test the ability of models to generalize to arbitrary image classification tasks in a zero-shot manner. It was not developed for general model deployment - to deploy models like CLIP, researchers will first need to carefully study their capabilities in relation to the specific context they’re being deployed within. The base model uses a ViT-L/14 Transformer architecture as an image encoder and uses a masked self-attention Transformer as a text encoder. These encoders are trained to maximize the similarity of (image, text) pairs via a contrastive loss. The original implementation had two variants: one using a ResNet image encoder and the other using a Vision Transformer. This repository has the variant with the Vision Transformer. The model is intended as a research output for research communities. We hope that this model will enable researchers to better understand and explore zero-shot, arbitrary image classification. We also hope it can be used for interdisciplinary studies of the potential impact of such models - the CLIP paper includes a discussion of potential downstream impacts to provide an example for this sort of analysis. The primary intended users of these models are AI researchers. We primarily imagine the model will be used by researchers to better understand robustness, generalization, and other capabilities, biases, and constraints of computer vision models. Any deployed use case of the model - whether commercial or not - is currently out of scope. Non-deployed use cases such as image search in a constrained environment, are also not recommended unless there is thorough in-domain testing of the model with a specific, fixed class taxonomy. This is because our safety assessment demonstrated a high need for task specific testing especially given the variability of CLIP’s performance with different class taxonomies. This makes untested and unconstrained deployment of the model in any use case currently potentially harmful. Certain use cases which would fall under the domain of surveillance and facial recognition are always out-of-scope regardless of performance of the model. This is because the use of artificial intelligence for tasks such as these can be premature currently given the lack of testing norms and checks to ensure its fair use. Since the model has not been purposefully trained in or evaluated on any languages other than English, its use should be limited to English language use cases. The model was trained on publicly available image-caption data. This was done through a combination of crawling a handful of websites and using commonly-used pre-existing image datasets such as YFCC100M. A large portion of the data comes from our crawling of the internet. This means that the data is more representative of people and societies most connected to the internet which tend to skew towards more developed nations, and younger, male users. Our goal with building this dataset was to test out robustness and generalizability in computer vision tasks. As a result, the focus was on gathering large quantities of data from different publicly-available internet data sources. The data was gathered in a mostly non-interventionist manner. However, we only crawled websites that had policies against excessively violent and adult images and allowed us to filter out such content. We do not intend for this dataset to be used as the basis for any commercial or deployed model and will not be releasing the dataset. We have evaluated the performance of CLIP on a wide range of benchmarks across a variety of computer vision datasets such as OCR to texture recognition to fine-grained classification. The paper describes model performance on the following datasets: - Food101 - CIFAR10 - CIFAR100 - Birdsnap - SUN397 - Stanford Cars - FGVC Aircraft - VOC2007 - DTD - Oxford-IIIT Pet dataset - Caltech101 - Flowers102 - MNIST - SVHN - IIIT5K - Hateful Memes - SST-2 - UCF101 - Kinetics700 - Country211 - CLEVR Counting - KITTI Distance - STL-10 - RareAct - Flickr30 - MSCOCO - ImageNet - ImageNet-A - ImageNet-R - ImageNet Sketch - ObjectNet (ImageNet Overlap) - Youtube-BB - ImageNet-Vid CLIP and our analysis of it have a number of limitations. CLIP currently struggles with respect to certain tasks such as fine grained classification and counting objects. CLIP also poses issues with regards to fairness and bias which we discuss in the paper and briefly in the next section. Additionally, our approach to testing CLIP also has an important limitation- in many cases we have used linear probes to evaluate the performance of CLIP and there is evidence suggesting that linear probes can underestimate model performance. We find that the performance of CLIP - and the specific biases it exhibits - can depend significantly on class design and the choices one makes for categories to include and exclude. We tested the risk of certain kinds of denigration with CLIP by classifying images of people from Fairface into crime-related and non-human animal categories. We found significant disparities with respect to race and gender. Additionally, we found that these disparities could shift based on how the classes were constructed. (Details captured in the Broader Impacts Section in the paper). We also tested the performance of CLIP on gender, race and age classification using the Fairface dataset (We default to using race categories as they are constructed in the Fairface dataset.) in order to assess quality of performance across different demographics. We found accuracy >96% across all races for gender classification with ‘Middle Eastern’ having the highest accuracy (98.4%) and ‘White’ having the lowest (96.5%). Additionally, CLIP averaged ~93% for racial classification and ~63% for age classification. Our use of evaluations to test for gender, race and age classification as well as denigration harms is simply to evaluate performance of the model across people and surface potential risks and not to demonstrate an endorsement/enthusiasm for such tasks. Where to send questions or comments about the model

Welcome to the gpt-oss series, OpenAI’s open-weight models designed for powerful reasoning, agentic tasks, and versatile developer use cases. We’re releasing two flavors of these open models: - `gpt-oss-120b` — for production, general purpose, high reasoning use cases that fit into a single 80GB GPU (like NVIDIA H100 or AMD MI300X) (117B parameters with 5.1B active parameters) - `gpt-oss-20b` — for lower latency, and local or specialized use cases (21B parameters with 3.6B active parameters) Both models were trained on our harmony response format and should only be used with the harmony format as it will not work correctly otherwise. > [!NOTE] > This model card is dedicated to the smaller `gpt-oss-20b` model. Check out `gpt-oss-120b` for the larger model. Permissive Apache 2.0 license: Build freely without copyleft restrictions or patent risk—ideal for experimentation, customization, and commercial deployment. Configurable reasoning effort: Easily adjust the reasoning effort (low, medium, high) based on your specific use case and latency needs. Full chain-of-thought: Gain complete access to the model’s reasoning process, facilitating easier debugging and increased trust in outputs. It’s not intended to be shown to end users. Fine-tunable: Fully customize models to your specific use case through parameter fine-tuning. Agentic capabilities: Use the models’ native capabilities for function calling, web browsing, Python code execution, and Structured Outputs. MXFP4 quantization: The models were post-trained with MXFP4 quantization of the MoE weights, making `gpt-oss-120b` run on a single 80GB GPU (like NVIDIA H100 or AMD MI300X) and the `gpt-oss-20b` model run within 16GB of memory. All evals were performed with the same MXFP4 quantization. You can use `gpt-oss-120b` and `gpt-oss-20b` with Transformers. If you use the Transformers chat template, it will automatically apply the harmony response format. If you use `model.generate` directly, you need to apply the harmony format manually using the chat template or use our openai-harmony package. To get started, install the necessary dependencies to setup your environment: Once, setup you can proceed to run the model by running the snippet below: Alternatively, you can run the model via `Transformers Serve` to spin up a OpenAI-compatible webserver: Learn more about how to use gpt-oss with Transformers. vLLM recommends using uv for Python dependency management. You can use vLLM to spin up an OpenAI-compatible webserver. The following command will automatically download the model and start the server. To learn about how to use this model with PyTorch and Triton, check out our reference implementations in the gpt-oss repository. If you are trying to run gpt-oss on consumer hardware, you can use Ollama by running the following commands after installing Ollama. If you are using LM Studio you can use the following commands to download. Check out our awesome list for a broader collection of gpt-oss resources and inference partners. You can download the model weights from the Hugging Face Hub directly from Hugging Face CLI: You can adjust the reasoning level that suits your task across three levels: Low: Fast responses for general dialogue. Medium: Balanced speed and detail. High: Deep and detailed analysis. The reasoning level can be set in the system prompts, e.g., "Reasoning: high". The gpt-oss models are excellent for: Web browsing (using built-in browsing tools) Function calling with defined schemas Agentic operations like browser tasks Both gpt-oss models can be fine-tuned for a variety of specialized use cases. This smaller model `gpt-oss-20b` can be fine-tuned on consumer hardware, whereas the larger `gpt-oss-120b` can be fine-tuned on a single H100 node.

Whisper is a state-of-the-art model for automatic speech recognition (ASR) and speech translation, proposed in the paper Robust Speech Recognition via Large-Scale Weak Supervision by Alec Radford et al. from OpenAI. Trained on >5M hours of labeled data, Whisper demonstrates a strong ability to generalise to many datasets and domains in a zero-shot setting. Whisper large-v3 has the same architecture as the previous large and large-v2 models, except for the following minor differences: 1. The spectrogram input uses 128 Mel frequency bins instead of 80 2. A new language token for Cantonese The Whisper large-v3 model was trained on 1 million hours of weakly labeled audio and 4 million hours of pseudo-labeled audio collected using Whisper large-v2 . The model was trained for 2.0 epochs over this mixture dataset. The large-v3 model shows improved performance over a wide variety of languages, showing 10% to 20% reduction of errors compared to Whisper large-v2 . For more details on the different checkpoints available, refer to the section Model details. Disclaimer: Content for this model card has partly been written by the 🤗 Hugging Face team, and partly copied and pasted from the original model card. Whisper large-v3 is supported in Hugging Face 🤗 Transformers. To run the model, first install the Transformers library. For this example, we'll also install 🤗 Datasets to load toy audio dataset from the Hugging Face Hub, and 🤗 Accelerate to reduce the model loading time: The model can be used with the `pipeline` class to transcribe audios of arbitrary length: To transcribe a local audio file, simply pass the path to your audio file when you call the pipeline: Multiple audio files can be transcribed in parallel by specifying them as a list and setting the `batchsize` parameter: Transformers is compatible with all Whisper decoding strategies, such as temperature fallback and condition on previous tokens. The following example demonstrates how to enable these heuristics: Whisper predicts the language of the source audio automatically. If the source audio language is known a-priori, it can be passed as an argument to the pipeline: By default, Whisper performs the task of speech transcription, where the source audio language is the same as the target text language. To perform speech translation, where the target text is in English, set the task to `"translate"`: Finally, the model can be made to predict timestamps. For sentence-level timestamps, pass the `returntimestamps` argument: The above arguments can be used in isolation or in combination. For example, to perform the task of speech transcription where the source audio is in French, and we want to return sentence-level timestamps, the following can be used: For more control over the generation parameters, use the model + processor API directly: You can apply additional speed and memory improvements to Whisper to further reduce the inference speed and VRAM requirements. Whisper has a receptive field of 30-seconds. To transcribe audios longer than this, one of two long-form algorithms are required: 1. Sequential: uses a "sliding window" for buffered inference, transcribing 30-second slices one after the other 2. Chunked: splits long audio files into shorter ones (with a small overlap between segments), transcribes each segment independently, and stitches the resulting transcriptions at the boundaries The sequential long-form algorithm should be used in either of the following scenarios: 1. Transcription accuracy is the most important factor, and speed is less of a consideration 2. You are transcribing batches of long audio files, in which case the latency of sequential is comparable to chunked, while being up to 0.5% WER more accurate Conversely, the chunked algorithm should be used when: 1. Transcription speed is the most important factor 2. You are transcribing a single long audio file By default, Transformers uses the sequential algorithm. To enable the chunked algorithm, pass the `chunklengths` parameter to the `pipeline`. For large-v3, a chunk length of 30-seconds is optimal. To activate batching over long audio files, pass the argument `batchsize`: The Whisper forward pass is compatible with `torch.compile` for 4.5x speed-ups. Note: `torch.compile` is currently not compatible with the Chunked long-form algorithm or Flash Attention 2 ⚠️ We recommend using Flash-Attention 2 if your GPU supports it and you are not using torch.compile. To do so, first install Flash Attention: Then pass `attnimplementation="flashattention2"` to `frompretrained`: If your GPU does not support Flash Attention, we recommend making use of PyTorch scaled dot-product attention (SDPA). This attention implementation is activated by default for PyTorch versions 2.1.1 or greater. To check whether you have a compatible PyTorch version, run the following Python code snippet: If the above returns `True`, you have a valid version of PyTorch installed and SDPA is activated by default. If it returns `False`, you need to upgrade your PyTorch version according to the official instructions Once a valid PyTorch version is installed, SDPA is activated by default. It can also be set explicitly by specifying `attnimplementation="sdpa"` as follows: For more information about how to use the SDPA refer to the Transformers SDPA documentation. Whisper is a Transformer based encoder-decoder model, also referred to as a sequence-to-sequence model. There are two flavours of Whisper model: English-only and multilingual. The English-only models were trained on the task of English speech recognition. The multilingual models were trained simultaneously on multilingual speech recognition and speech translation. For speech recognition, the model predicts transcriptions in the same language as the audio. For speech translation, the model predicts transcriptions to a different language to the audio. Whisper checkpoints come in five configurations of varying model sizes. The smallest four are available as English-only and multilingual. The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints are available on the Hugging Face Hub. The checkpoints are summarised in the following table with links to the models on the Hub: | Size | Parameters | English-only | Multilingual | |----------|------------|------------------------------------------------------|-----------------------------------------------------| | tiny | 39 M | ✓ | ✓ | | base | 74 M | ✓ | ✓ | | small | 244 M | ✓ | ✓ | | medium | 769 M | ✓ | ✓ | | large | 1550 M | x | ✓ | | large-v2 | 1550 M | x | ✓ | | large-v3 | 1550 M | x | ✓ | The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However, its predictive capabilities can be improved further for certain languages and tasks through fine-tuning. The blog post Fine-Tune Whisper with 🤗 Transformers provides a step-by-step guide to fine-tuning the Whisper model with as little as 5 hours of labelled data. The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research. The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them. In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes. The large-v3 checkpoint is trained on 1 million hours of weakly labeled audio and 4 million hours of pseudo-labeled audio collected using Whisper large-v2. As discussed in the accompanying paper, we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language. Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level. However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself. Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in the paper accompanying this release. In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in the paper. It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages. We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications. There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.

Whisper is a state-of-the-art model for automatic speech recognition (ASR) and speech translation, proposed in the paper Robust Speech Recognition via Large-Scale Weak Supervision by Alec Radford et al. from OpenAI. Trained on >5M hours of labeled data, Whisper demonstrates a strong ability to generalise to many datasets and domains in a zero-shot setting. Whisper large-v3-turbo is a finetuned version of a pruned Whisper large-v3. In other words, it's the exact same model, except that the number of decoding layers have reduced from 32 to 4. As a result, the model is way faster, at the expense of a minor quality degradation. You can find more details about it in this GitHub discussion. Disclaimer: Content for this model card has partly been written by the 🤗 Hugging Face team, and partly copied and pasted from the original model card. Whisper large-v3-turbo is supported in Hugging Face 🤗 Transformers. To run the model, first install the Transformers library. For this example, we'll also install 🤗 Datasets to load toy audio dataset from the Hugging Face Hub, and 🤗 Accelerate to reduce the model loading time: The model can be used with the `pipeline` class to transcribe audios of arbitrary length: To transcribe a local audio file, simply pass the path to your audio file when you call the pipeline: Multiple audio files can be transcribed in parallel by specifying them as a list and setting the `batchsize` parameter: Transformers is compatible with all Whisper decoding strategies, such as temperature fallback and condition on previous tokens. The following example demonstrates how to enable these heuristics: Whisper predicts the language of the source audio automatically. If the source audio language is known a-priori, it can be passed as an argument to the pipeline: By default, Whisper performs the task of speech transcription, where the source audio language is the same as the target text language. To perform speech translation, where the target text is in English, set the task to `"translate"`: Finally, the model can be made to predict timestamps. For sentence-level timestamps, pass the `returntimestamps` argument: The above arguments can be used in isolation or in combination. For example, to perform the task of speech transcription where the source audio is in French, and we want to return sentence-level timestamps, the following can be used: For more control over the generation parameters, use the model + processor API directly: You can apply additional speed and memory improvements to Whisper to further reduce the inference speed and VRAM requirements. Whisper has a receptive field of 30-seconds. To transcribe audios longer than this, one of two long-form algorithms are required: 1. Sequential: uses a "sliding window" for buffered inference, transcribing 30-second slices one after the other 2. Chunked: splits long audio files into shorter ones (with a small overlap between segments), transcribes each segment independently, and stitches the resulting transcriptions at the boundaries The sequential long-form algorithm should be used in either of the following scenarios: 1. Transcription accuracy is the most important factor, and speed is less of a consideration 2. You are transcribing batches of long audio files, in which case the latency of sequential is comparable to chunked, while being up to 0.5% WER more accurate Conversely, the chunked algorithm should be used when: 1. Transcription speed is the most important factor 2. You are transcribing a single long audio file By default, Transformers uses the sequential algorithm. To enable the chunked algorithm, pass the `chunklengths` parameter to the `pipeline`. For large-v3, a chunk length of 30-seconds is optimal. To activate batching over long audio files, pass the argument `batchsize`: The Whisper forward pass is compatible with `torch.compile` for 4.5x speed-ups. Note: `torch.compile` is currently not compatible with the Chunked long-form algorithm or Flash Attention 2 ⚠️ We recommend using Flash-Attention 2 if your GPU supports it and you are not using torch.compile. To do so, first install Flash Attention: Then pass `attnimplementation="flashattention2"` to `frompretrained`: If your GPU does not support Flash Attention, we recommend making use of PyTorch scaled dot-product attention (SDPA). This attention implementation is activated by default for PyTorch versions 2.1.1 or greater. To check whether you have a compatible PyTorch version, run the following Python code snippet: If the above returns `True`, you have a valid version of PyTorch installed and SDPA is activated by default. If it returns `False`, you need to upgrade your PyTorch version according to the official instructions Once a valid PyTorch version is installed, SDPA is activated by default. It can also be set explicitly by specifying `attnimplementation="sdpa"` as follows: For more information about how to use the SDPA refer to the Transformers SDPA documentation. Whisper is a Transformer based encoder-decoder model, also referred to as a sequence-to-sequence model. There are two flavours of Whisper model: English-only and multilingual. The English-only models were trained on the task of English speech recognition. The multilingual models were trained simultaneously on multilingual speech recognition and speech translation. For speech recognition, the model predicts transcriptions in the same language as the audio. For speech translation, the model predicts transcriptions to a different language to the audio. Whisper checkpoints come in five configurations of varying model sizes. The smallest four are available as English-only and multilingual. The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints are available on the Hugging Face Hub. The checkpoints are summarised in the following table with links to the models on the Hub: | Size | Parameters | English-only | Multilingual | |----------|------------|------------------------------------------------------|-----------------------------------------------------| | tiny | 39 M | ✓ | ✓ | | base | 74 M | ✓ | ✓ | | small | 244 M | ✓ | ✓ | | medium | 769 M | ✓ | ✓ | | large | 1550 M | x | ✓ | | large-v2 | 1550 M | x | ✓ | | large-v3 | 1550 M | x | ✓ | | large-v3-turbo | 809 M | x | ✓ | The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However, its predictive capabilities can be improved further for certain languages and tasks through fine-tuning. The blog post Fine-Tune Whisper with 🤗 Transformers provides a step-by-step guide to fine-tuning the Whisper model with as little as 5 hours of labelled data. The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research. The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them. In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes. Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level. However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself. Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in the paper accompanying this release. In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in the paper. It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages. We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications. There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.

Welcome to the gpt-oss series, OpenAI’s open-weight models designed for powerful reasoning, agentic tasks, and versatile developer use cases. We’re releasing two flavors of these open models: - `gpt-oss-120b` — for production, general purpose, high reasoning use cases that fit into a single 80GB GPU (like NVIDIA H100 or AMD MI300X) (117B parameters with 5.1B active parameters) - `gpt-oss-20b` — for lower latency, and local or specialized use cases (21B parameters with 3.6B active parameters) Both models were trained on our harmony response format and should only be used with the harmony format as it will not work correctly otherwise. > [!NOTE] > This model card is dedicated to the larger `gpt-oss-120b` model. Check out `gpt-oss-20b` for the smaller model. Permissive Apache 2.0 license: Build freely without copyleft restrictions or patent risk—ideal for experimentation, customization, and commercial deployment. Configurable reasoning effort: Easily adjust the reasoning effort (low, medium, high) based on your specific use case and latency needs. Full chain-of-thought: Gain complete access to the model’s reasoning process, facilitating easier debugging and increased trust in outputs. It’s not intended to be shown to end users. Fine-tunable: Fully customize models to your specific use case through parameter fine-tuning. Agentic capabilities: Use the models’ native capabilities for function calling, web browsing, Python code execution, and Structured Outputs. MXFP4 quantization: The models were post-trained with MXFP4 quantization of the MoE weights, making `gpt-oss-120b` run on a single 80GB GPU (like NVIDIA H100 or AMD MI300X) and the `gpt-oss-20b` model run within 16GB of memory. All evals were performed with the same MXFP4 quantization. You can use `gpt-oss-120b` and `gpt-oss-20b` with Transformers. If you use the Transformers chat template, it will automatically apply the harmony response format. If you use `model.generate` directly, you need to apply the harmony format manually using the chat template or use our openai-harmony package. To get started, install the necessary dependencies to setup your environment: Once, setup you can proceed to run the model by running the snippet below: Alternatively, you can run the model via `Transformers Serve` to spin up a OpenAI-compatible webserver: Learn more about how to use gpt-oss with Transformers. vLLM recommends using uv for Python dependency management. You can use vLLM to spin up an OpenAI-compatible webserver. The following command will automatically download the model and start the server. To learn about how to use this model with PyTorch and Triton, check out our reference implementations in the gpt-oss repository. If you are trying to run gpt-oss on consumer hardware, you can use Ollama by running the following commands after installing Ollama. If you are using LM Studio you can use the following commands to download. Check out our awesome list for a broader collection of gpt-oss resources and inference partners. You can download the model weights from the Hugging Face Hub directly from Hugging Face CLI: You can adjust the reasoning level that suits your task across three levels: Low: Fast responses for general dialogue. Medium: Balanced speed and detail. High: Deep and detailed analysis. The reasoning level can be set in the system prompts, e.g., "Reasoning: high". The gpt-oss models are excellent for: Web browsing (using built-in browsing tools) Function calling with defined schemas Agentic operations like browser tasks Both gpt-oss models can be fine-tuned for a variety of specialized use cases. This larger model `gpt-oss-120b` can be fine-tuned on a single H100 node, whereas the smaller `gpt-oss-20b` can even be fine-tuned on consumer hardware.

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - optimizer: None - trainingprecision: float32 - Transformers 4.21.3 - TensorFlow 2.8.2 - Tokenizers 0.12.1

--- tags: - vision widget: - src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/cat-dog-music.png candidate_labels: playing music, playing sports example_title: Cat & Dog ---

--- language: - en - zh - de - es - ru - ko - fr - ja - pt - tr - pl - ca - nl - ar - sv - it - id - hi - fi - vi - he - uk - el - ms - cs - ro - da - hu - ta - no - th - ur - hr - bg - lt - la - mi - ml - cy - sk - te - fa - lv - bn - sr - az - sl - kn - et - mk - br - eu - is - hy - ne - mn - bs - kk - sq - sw - gl - mr - pa - si - km - sn - yo - so - af - oc - ka - be - tg - sd - gu - am - yi - lo - uz - fo - ht - ps - tk - nn - mt - sa - lb - my - bo - tl - mg - as - tt - haw - ln - ha - ba

--- language: - en - zh - de - es - ru - ko - fr - ja - pt - tr - pl - ca - nl - ar - sv - it - id - hi - fi - vi - he - uk - el - ms - cs - ro - da - hu - ta - no - th - ur - hr - bg - lt - la - mi - ml - cy - sk - te - fa - lv - bn - sr - az - sl - kn - et - mk - br - eu - is - hy - ne - mn - bs - kk - sq - sw - gl - mr - pa - si - km - sn - yo - so - af - oc - ka - be - tg - sd - gu - am - yi - lo - uz - fo - ht - ps - tk - nn - mt - sa - lb - my - bo - tl - mg - as - tt - haw - ln - ha - ba

--- language: - en - zh - de - es - ru - ko - fr - ja - pt - tr - pl - ca - nl - ar - sv - it - id - hi - fi - vi - he - uk - el - ms - cs - ro - da - hu - ta - no - th - ur - hr - bg - lt - la - mi - ml - cy - sk - te - fa - lv - bn - sr - az - sl - kn - et - mk - br - eu - is - hy - ne - mn - bs - kk - sq - sw - gl - mr - pa - si - km - sn - yo - so - af - oc - ka - be - tg - sd - gu - am - yi - lo - uz - fo - ht - ps - tk - nn - mt - sa - lb - my - bo - tl - mg - as - tt - haw - ln - ha - ba

--- language: - en - zh - de - es - ru - ko - fr - ja - pt - tr - pl - ca - nl - ar - sv - it - id - hi - fi - vi - he - uk - el - ms - cs - ro - da - hu - ta - no - th - ur - hr - bg - lt - la - mi - ml - cy - sk - te - fa - lv - bn - sr - az - sl - kn - et - mk - br - eu - is - hy - ne - mn - bs - kk - sq - sw - gl - mr - pa - si - km - sn - yo - so - af - oc - ka - be - tg - sd - gu - am - yi - lo - uz - fo - ht - ps - tk - nn - mt - sa - lb - my - bo - tl - mg - as - tt - haw - ln - ha - ba

--- language: - en - zh - de - es - ru - ko - fr - ja - pt - tr - pl - ca - nl - ar - sv - it - id - hi - fi - vi - he - uk - el - ms - cs - ro - da - hu - ta - no - th - ur - hr - bg - lt - la - mi - ml - cy - sk - te - fa - lv - bn - sr - az - sl - kn - et - mk - br - eu - is - hy - ne - mn - bs - kk - sq - sw - gl - mr - pa - si - km - sn - yo - so - af - oc - ka - be - tg - sd - gu - am - yi - lo - uz - fo - ht - ps - tk - nn - mt - sa - lb - my - bo - tl - mg - as - tt - haw - ln - ha - ba

Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation. Trained on 680k hours of labelled data, Whisper models demonstrate a strong ability to generalise to many datasets and domains without the need for fine-tuning. Whisper was proposed in the paper Robust Speech Recognition via Large-Scale Weak Supervision by Alec Radford et al. from OpenAI. The original code repository can be found here. Disclaimer: Content for this model card has partly been written by the Hugging Face team, and parts of it were copied and pasted from the original model card. Whisper is a Transformer based encoder-decoder model, also referred to as a sequence-to-sequence model. It was trained on 680k hours of labelled speech data annotated using large-scale weak supervision. The models were trained on either English-only data or multilingual data. The English-only models were trained on the task of speech recognition. The multilingual models were trained on both speech recognition and speech translation. For speech recognition, the model predicts transcriptions in the same language as the audio. For speech translation, the model predicts transcriptions to a different language to the audio. Whisper checkpoints come in five configurations of varying model sizes. The smallest four are trained on either English-only or multilingual data. The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints are available on the Hugging Face Hub. The checkpoints are summarised in the following table with links to the models on the Hub: | Size | Parameters | English-only | Multilingual | |----------|------------|------------------------------------------------------|-----------------------------------------------------| | tiny | 39 M | ✓ | ✓ | | base | 74 M | ✓ | ✓ | | small | 244 M | ✓ | ✓ | | medium | 769 M | ✓ | ✓ | | large | 1550 M | x | ✓ | | large-v2 | 1550 M | x | ✓ | This checkpoint is an English-only model, meaning it can be used for English speech recognition. Multilingual speech recognition or speech translation is possible through use of a multilingual checkpoint. To transcribe audio samples, the model has to be used alongside a `WhisperProcessor`. The `WhisperProcessor` is used to: 1. Pre-process the audio inputs (converting them to log-Mel spectrograms for the model) 2. Post-process the model outputs (converting them from tokens to text) The context tokens can be removed from the start of the transcription by setting `skipspecialtokens=True`. This code snippet shows how to evaluate Whisper tiny.en on LibriSpeech test-clean: The Whisper model is intrinsically designed to work on audio samples of up to 30s in duration. However, by using a chunking algorithm, it can be used to transcribe audio samples of up to arbitrary length. This is possible through Transformers `pipeline` method. Chunking is enabled by setting `chunklengths=30` when instantiating the pipeline. With chunking enabled, the pipeline can be run with batched inference. It can also be extended to predict sequence level timestamps by passing `returntimestamps=True`: Refer to the blog post ASR Chunking for more details on the chunking algorithm. The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However, its predictive capabilities can be improved further for certain languages and tasks through fine-tuning. The blog post Fine-Tune Whisper with 🤗 Transformers provides a step-by-step guide to fine-tuning the Whisper model with as little as 5 hours of labelled data. The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research. The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them. In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes. The models are trained on 680,000 hours of audio and the corresponding transcripts collected from the internet. 65% of this data (or 438,000 hours) represents English-language audio and matched English transcripts, roughly 18% (or 126,000 hours) represents non-English audio and English transcripts, while the final 17% (or 117,000 hours) represents non-English audio and the corresponding transcript. This non-English data represents 98 different languages. As discussed in the accompanying paper, we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language. Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level. However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself. Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in the paper accompanying this release. In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in the paper. It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages. We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications. There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.

Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation. Trained on 680k hours of labelled data, Whisper models demonstrate a strong ability to generalise to many datasets and domains without the need for fine-tuning. Whisper was proposed in the paper Robust Speech Recognition via Large-Scale Weak Supervision by Alec Radford et al. from OpenAI. The original code repository can be found here. Disclaimer: Content for this model card has partly been written by the Hugging Face team, and parts of it were copied and pasted from the original model card. Whisper is a Transformer based encoder-decoder model, also referred to as a sequence-to-sequence model. It was trained on 680k hours of labelled speech data annotated using large-scale weak supervision. The models were trained on either English-only data or multilingual data. The English-only models were trained on the task of speech recognition. The multilingual models were trained on both speech recognition and speech translation. For speech recognition, the model predicts transcriptions in the same language as the audio. For speech translation, the model predicts transcriptions to a different language to the audio. Whisper checkpoints come in five configurations of varying model sizes. The smallest four are trained on either English-only or multilingual data. The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints are available on the Hugging Face Hub. The checkpoints are summarised in the following table with links to the models on the Hub: | Size | Parameters | English-only | Multilingual | |----------|------------|------------------------------------------------------|-----------------------------------------------------| | tiny | 39 M | ✓ | ✓ | | base | 74 M | ✓ | ✓ | | small | 244 M | ✓ | ✓ | | medium | 769 M | ✓ | ✓ | | large | 1550 M | x | ✓ | | large-v2 | 1550 M | x | ✓ | This checkpoint is an English-only model, meaning it can be used for English speech recognition. Multilingual speech recognition or speech translation is possible through use of a multilingual checkpoint. To transcribe audio samples, the model has to be used alongside a `WhisperProcessor`. The `WhisperProcessor` is used to: 1. Pre-process the audio inputs (converting them to log-Mel spectrograms for the model) 2. Post-process the model outputs (converting them from tokens to text) The context tokens can be removed from the start of the transcription by setting `skipspecialtokens=True`. This code snippet shows how to evaluate Whisper medium.en on LibriSpeech test-clean: The Whisper model is intrinsically designed to work on audio samples of up to 30s in duration. However, by using a chunking algorithm, it can be used to transcribe audio samples of up to arbitrary length. This is possible through Transformers `pipeline` method. Chunking is enabled by setting `chunklengths=30` when instantiating the pipeline. With chunking enabled, the pipeline can be run with batched inference. It can also be extended to predict sequence level timestamps by passing `returntimestamps=True`: Refer to the blog post ASR Chunking for more details on the chunking algorithm. Fine-Tuning The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However, its predictive capabilities can be improved further for certain languages and tasks through fine-tuning. The blog post Fine-Tune Whisper with 🤗 Transformers provides a step-by-step guide to fine-tuning the Whisper model with as little as 5 hours of labelled data. The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research. The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them. In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes. The models are trained on 680,000 hours of audio and the corresponding transcripts collected from the internet. 65% of this data (or 438,000 hours) represents English-language audio and matched English transcripts, roughly 18% (or 126,000 hours) represents non-English audio and English transcripts, while the final 17% (or 117,000 hours) represents non-English audio and the corresponding transcript. This non-English data represents 98 different languages. As discussed in the accompanying paper, we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language. Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level. However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself. Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in the paper accompanying this release. In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in the paper. It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages. We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications. There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.

Whisper is a pre-trained model for automatic speech recognition (ASR) and speech translation. Trained on 680k hours of labelled data, Whisper models demonstrate a strong ability to generalise to many datasets and domains without the need for fine-tuning. Whisper was proposed in the paper Robust Speech Recognition via Large-Scale Weak Supervision by Alec Radford et al. from OpenAI. The original code repository can be found here. Disclaimer: Content for this model card has partly been written by the Hugging Face team, and parts of it were copied and pasted from the original model card. Whisper is a Transformer based encoder-decoder model, also referred to as a sequence-to-sequence model. It was trained on 680k hours of labelled speech data annotated using large-scale weak supervision. The models were trained on either English-only data or multilingual data. The English-only models were trained on the task of speech recognition. The multilingual models were trained on both speech recognition and speech translation. For speech recognition, the model predicts transcriptions in the same language as the audio. For speech translation, the model predicts transcriptions to a different language to the audio. Whisper checkpoints come in five configurations of varying model sizes. The smallest four are trained on either English-only or multilingual data. The largest checkpoints are multilingual only. All ten of the pre-trained checkpoints are available on the Hugging Face Hub. The checkpoints are summarised in the following table with links to the models on the Hub: | Size | Parameters | English-only | Multilingual | |----------|------------|------------------------------------------------------|-----------------------------------------------------| | tiny | 39 M | ✓ | ✓ | | base | 74 M | ✓ | ✓ | | small | 244 M | ✓ | ✓ | | medium | 769 M | ✓ | ✓ | | large | 1550 M | x | ✓ | | large-v2 | 1550 M | x | ✓ | This checkpoint is an English-only model, meaning it can be used for English speech recognition. Multilingual speech recognition or speech translation is possible through use of a multilingual checkpoint. To transcribe audio samples, the model has to be used alongside a `WhisperProcessor`. The `WhisperProcessor` is used to: 1. Pre-process the audio inputs (converting them to log-Mel spectrograms for the model) 2. Post-process the model outputs (converting them from tokens to text) The context tokens can be removed from the start of the transcription by setting `skipspecialtokens=True`. This code snippet shows how to evaluate Whisper base.en on LibriSpeech test-clean: The Whisper model is intrinsically designed to work on audio samples of up to 30s in duration. However, by using a chunking algorithm, it can be used to transcribe audio samples of up to arbitrary length. This is possible through Transformers `pipeline` method. Chunking is enabled by setting `chunklengths=30` when instantiating the pipeline. With chunking enabled, the pipeline can be run with batched inference. It can also be extended to predict sequence level timestamps by passing `returntimestamps=True`: Refer to the blog post ASR Chunking for more details on the chunking algorithm. The pre-trained Whisper model demonstrates a strong ability to generalise to different datasets and domains. However, its predictive capabilities can be improved further for certain languages and tasks through fine-tuning. The blog post Fine-Tune Whisper with 🤗 Transformers provides a step-by-step guide to fine-tuning the Whisper model with as little as 5 hours of labelled data. The primary intended users of these models are AI researchers studying robustness, generalization, capabilities, biases, and constraints of the current model. However, Whisper is also potentially quite useful as an ASR solution for developers, especially for English speech recognition. We recognize that once models are released, it is impossible to restrict access to only “intended” uses or to draw reasonable guidelines around what is or is not research. The models are primarily trained and evaluated on ASR and speech translation to English tasks. They show strong ASR results in ~10 languages. They may exhibit additional capabilities, particularly if fine-tuned on certain tasks like voice activity detection, speaker classification, or speaker diarization but have not been robustly evaluated in these areas. We strongly recommend that users perform robust evaluations of the models in a particular context and domain before deploying them. In particular, we caution against using Whisper models to transcribe recordings of individuals taken without their consent or purporting to use these models for any kind of subjective classification. We recommend against use in high-risk domains like decision-making contexts, where flaws in accuracy can lead to pronounced flaws in outcomes. The models are intended to transcribe and translate speech, use of the model for classification is not only not evaluated but also not appropriate, particularly to infer human attributes. The models are trained on 680,000 hours of audio and the corresponding transcripts collected from the internet. 65% of this data (or 438,000 hours) represents English-language audio and matched English transcripts, roughly 18% (or 126,000 hours) represents non-English audio and English transcripts, while the final 17% (or 117,000 hours) represents non-English audio and the corresponding transcript. This non-English data represents 98 different languages. As discussed in the accompanying paper, we see that performance on transcription in a given language is directly correlated with the amount of training data we employ in that language. Our studies show that, over many existing ASR systems, the models exhibit improved robustness to accents, background noise, technical language, as well as zero shot translation from multiple languages into English; and that accuracy on speech recognition and translation is near the state-of-the-art level. However, because the models are trained in a weakly supervised manner using large-scale noisy data, the predictions may include texts that are not actually spoken in the audio input (i.e. hallucination). We hypothesize that this happens because, given their general knowledge of language, the models combine trying to predict the next word in audio with trying to transcribe the audio itself. Our models perform unevenly across languages, and we observe lower accuracy on low-resource and/or low-discoverability languages or languages where we have less training data. The models also exhibit disparate performance on different accents and dialects of particular languages, which may include higher word error rate across speakers of different genders, races, ages, or other demographic criteria. Our full evaluation results are presented in the paper accompanying this release. In addition, the sequence-to-sequence architecture of the model makes it prone to generating repetitive texts, which can be mitigated to some degree by beam search and temperature scheduling but not perfectly. Further analysis on these limitations are provided in the paper. It is likely that this behavior and hallucinations may be worse on lower-resource and/or lower-discoverability languages. We anticipate that Whisper models’ transcription capabilities may be used for improving accessibility tools. While Whisper models cannot be used for real-time transcription out of the box – their speed and size suggest that others may be able to build applications on top of them that allow for near-real-time speech recognition and translation. The real value of beneficial applications built on top of Whisper models suggests that the disparate performance of these models may have real economic implications. There are also potential dual use concerns that come with releasing Whisper. While we hope the technology will be used primarily for beneficial purposes, making ASR technology more accessible could enable more actors to build capable surveillance technologies or scale up existing surveillance efforts, as the speed and accuracy allow for affordable automatic transcription and translation of large volumes of audio communication. Moreover, these models may have some capabilities to recognize specific individuals out of the box, which in turn presents safety concerns related both to dual use and disparate performance. In practice, we expect that the cost of transcription is not the limiting factor of scaling up surveillance projects.

Try gpt-oss-safeguard · Guide · Model card · OpenAI blog `gpt-oss-safeguard-120b` and `gpt-oss-safeguard-20b` are safety reasoning models built-upon gpt-oss. With these models, you can classify text content based on safety policies that you provide and perform a suite of foundational safety tasks. These models are intended for safety use cases. For other applications, we recommend using gpt-oss models. This model `gpt-oss-safeguard-120b` fits into a single H100 GPU (117B parameters with 5.1B active parameters). Check out `gpt-oss-safeguard-20b` for lower latency (21B parameters with 3.6B active parameters). Both models were trained on our harmony response format and should only be used with the harmony format as it will not work correctly otherwise. Trained to reason about safety : Trained and tuned for safety reasoning to accommodate use cases like LLM input-output filtering, online content labeling and offline labeling for Trust and Safety use cases. Bring your own policy: Interprets your written policy, so it generalizes across products and use cases with minimal engineering. Reasoned decisions, not just scores: Gain complete access to the model’s reasoning process, facilitating easier debugging and increased trust in policy decisions. Keep in mind Raw CoT is meant for developers and safety practitioners. It’s not intended for exposure to general users or use cases outside of safety contexts. Configurable reasoning effort: Easily adjust the reasoning effort (low, medium, high) based on your specific use case and latency needs. Permissive Apache 2.0 license: Build freely without copyleft restrictions or patent risk—ideal for experimentation, customization, and commercial deployment. You can use gpt-oss-safeguard-120b and gpt-oss-safeguard-20b similar to gpt-oss-120b and gpt-oss-20b as described in our respective cookbooks. We’ve also provided a detailed prompting guide that provides guidelines for how to craft your policy and use it with the models. To download the model weights from Hugging Face hub using similar instructions to gpt-oss-120b. gpt-oss-safeguard is a model partner of the Robust Open Online Safety Tools (ROOST) Model Community. The ROOST Model Community (RMC) is a group of safety practitioners exploring open source AI models to protect online spaces. As an RMC model partner, OpenAI is committed to incorporating user feedback and jointly iterating on future releases in pursuit of open safety. Visit the RMC GitHub repo to learn more about this partnership and how to get involved.

Try gpt-oss-safeguard · Guide · Model card · OpenAI blog `gpt-oss-safeguard-120b` and `gpt-oss-safeguard-20b` are safety reasoning models built-upon gpt-oss. With these models, you can classify text content based on safety policies that you provide and perform a suite of foundational safety tasks. These models are intended for safety use cases. For other applications, we recommend using gpt-oss models. This model `gpt-oss-safeguard-20b` (21B parameters with 3.6B active parameters) fits into GPUs with 16GB of VRAM. Check out `gpt-oss-safeguard-120b` (117B parameters with 5.1B active parameters) for the larger model. Both models were trained on our harmony response format and should only be used with the harmony format as it will not work correctly otherwise. Trained to reason about safety : Trained and tuned for safety reasoning to accommodate use cases like LLM input-output filtering, online content labeling and offline labeling for Trust and Safety use cases. Bring your own policy: Interprets your written policy, so it generalizes across products and use cases with minimal engineering. Reasoned decisions, not just scores: Gain complete access to the model’s reasoning process, facilitating easier debugging and increased trust in policy decisions. Keep in mind Raw CoT is meant for developers and safety practitioners. It’s not intended for exposure to general users or use cases outside of safety contexts. Configurable reasoning effort: Easily adjust the reasoning effort (low, medium, high) based on your specific use case and latency needs. Permissive Apache 2.0 license: Build freely without copyleft restrictions or patent risk—ideal for experimentation, customization, and commercial deployment. You can use gpt-oss-safeguard-120b and gpt-oss-safeguard-20b similar to gpt-oss-120b and gpt-oss-20b as described in our respective cookbooks. We’ve also provided a detailed prompting guide that provides guidelines for how to craft your policy and use it with the models. To download the model weights from Hugging Face hub using similar instructions to gpt-oss-120b. gpt-oss-safeguard is a model partner of the Robust Open Online Safety Tools (ROOST) Model Community. The ROOST Model Community (RMC) is a group of safety practitioners exploring open source AI models to protect online spaces. As an RMC model partner, OpenAI is committed to incorporating user feedback and jointly iterating on future releases in pursuit of open safety. Visit the RMC GitHub repo to learn more about this partnership and how to get involved.

Shap-E introduces a diffusion process that can generate a 3D image from a text prompt. It was introduced in Shap-E: Generating Conditional 3D Implicit Functions by Heewoo Jun and Alex Nichol from OpenAI. Original repository of Shap-E can be found here: https://github.com/openai/shap-e. The authors of Shap-E didn't author this model card. They provide a separate model card here. We present Shap-E, a conditional generative model for 3D assets. Unlike recent work on 3D generative models which produce a single output representation, Shap-E directly generates the parameters of implicit functions that can be rendered as both textured meshes and neural radiance fields. We train Shap-E in two stages: first, we train an encoder that deterministically maps 3D assets into the parameters of an implicit function; second, we train a conditional diffusion model on outputs of the encoder. When trained on a large dataset of paired 3D and text data, our resulting models are capable of generating complex and diverse 3D assets in a matter of seconds. When compared to Point-E, an explicit generative model over point clouds, Shap-E converges faster and reaches comparable or better sample quality despite modeling a higher-dimensional, multi-representation output space. We release model weights, inference code, and samples at this https URL. openai/shap-e: produces a 3D image from a text input prompt openai/shap-e-img2img: samples a 3D image from synthetic 2D image First make sure you have installed all the dependencies: Once the dependencies are installed, use the code below: