Text Generation
Transformers
PyTorch
Safetensors
English
gpt_neox
causal-lm
pythia
text-generation-inference
pythia-12b / EleutherAI_pythia-12b.json
RiccardoDav's picture
add AIBOM
56ce20c verified
raw
history blame
7.97 kB
{
"bomFormat": "CycloneDX",
"specVersion": "1.6",
"serialNumber": "urn:uuid:e33947f8-aae7-4995-adeb-d36d728818aa",
"version": 1,
"metadata": {
"timestamp": "2025-10-07T08:19:30.409870+00:00",
"component": {
"type": "machine-learning-model",
"bom-ref": "EleutherAI/pythia-12b-e54a9a31-ff47-5eb1-8c49-d06c73636dc7",
"licenses": [
{
"license": {
"id": "Apache-2.0",
"url": "https://spdx.org/licenses/Apache-2.0.html"
}
}
],
"externalReferences": [
{
"url": "https://huggingface.co/EleutherAI/pythia-12b",
"type": "documentation"
}
],
"modelCard": {
"modelParameters": {
"datasets": [
{
"ref": "EleutherAI/pile-671dd665-c5ab-5ddc-8ee4-6f2955b81787"
}
],
"task": "text-generation",
"architectureFamily": "gpt_neox",
"modelArchitecture": "GPTNeoXForCausalLM"
},
"properties": [
{
"name": "library_name",
"value": "transformers"
}
],
"consideration": {
"useCases": "## Uses and Limitations\n"
}
},
"name": "EleutherAI/pythia-12b",
"authors": [
{
"name": "EleutherAI"
}
],
"description": "## Model Details\n\n- Developed by: [EleutherAI](http://eleuther.ai)\n- Model type: Transformer-based Language Model\n- Language: English\n- Learn more: [Pythia's GitHub repository](https://github.com/EleutherAI/pythia)\nfor training procedure, config files, and details on how to use.\n[See paper](https://arxiv.org/pdf/2304.01373.pdf) for more evals and implementation\ndetails.\n- Library: [GPT-NeoX](https://github.com/EleutherAI/gpt-neox)\n- License: Apache 2.0\n- Contact: to ask questions about this model, join the [EleutherAI\nDiscord](https://discord.gg/zBGx3azzUn), and post them in `#release-discussion`.\nPlease read the existing *Pythia* documentation before asking about it in the\nEleutherAI Discord. For general correspondence: [contact@eleuther.\nai](mailto:contact@eleuther.ai).\n\n<figure>\n\n| Pythia model | Non-Embedding Params | Layers | Model Dim | Heads | Batch Size | Learning Rate | Equivalent Models |\n| -----------: | -------------------: | :----: | :-------: | :---: | :--------: | :-------------------: | :--------------------: |\n| 70M | 18,915,328 | 6 | 512 | 8 | 2M | 1.0 x 10<sup>-3</sup> | \u2014 |\n| 160M | 85,056,000 | 12 | 768 | 12 | 2M | 6.0 x 10<sup>-4</sup> | GPT-Neo 125M, OPT-125M |\n| 410M | 302,311,424 | 24 | 1024 | 16 | 2M | 3.0 x 10<sup>-4</sup> | OPT-350M |\n| 1.0B | 805,736,448 | 16 | 2048 | 8 | 2M | 3.0 x 10<sup>-4</sup> | \u2014 |\n| 1.4B | 1,208,602,624 | 24 | 2048 | 16 | 2M | 2.0 x 10<sup>-4</sup> | GPT-Neo 1.3B, OPT-1.3B |\n| 2.8B | 2,517,652,480 | 32 | 2560 | 32 | 2M | 1.6 x 10<sup>-4</sup> | GPT-Neo 2.7B, OPT-2.7B |\n| 6.9B | 6,444,163,072 | 32 | 4096 | 32 | 2M | 1.2 x 10<sup>-4</sup> | OPT-6.7B |\n| 12B | 11,327,027,200 | 36 | 5120 | 40 | 2M | 1.2 x 10<sup>-4</sup> | \u2014 |\n<figcaption>Engineering details for the <i>Pythia Suite</i>. Deduped and\nnon-deduped models of a given size have the same hyperparameters. \u201cEquivalent\u201d\nmodels have <b>exactly</b> the same architecture, and the same number of\nnon-embedding parameters.</figcaption>\n</figure>\n",
"tags": [
"transformers",
"pytorch",
"safetensors",
"gpt_neox",
"text-generation",
"causal-lm",
"pythia",
"en",
"dataset:EleutherAI/pile",
"arxiv:2304.01373",
"arxiv:2101.00027",
"arxiv:2201.07311",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
]
}
},
"components": [
{
"type": "data",
"bom-ref": "EleutherAI/pile-671dd665-c5ab-5ddc-8ee4-6f2955b81787",
"name": "EleutherAI/pile",
"data": [
{
"type": "dataset",
"bom-ref": "EleutherAI/pile-671dd665-c5ab-5ddc-8ee4-6f2955b81787",
"name": "EleutherAI/pile",
"contents": {
"url": "https://huggingface.co/datasets/EleutherAI/pile",
"properties": [
{
"name": "task_categories",
"value": "text-generation, fill-mask"
},
{
"name": "task_ids",
"value": "language-modeling, masked-language-modeling"
},
{
"name": "language",
"value": "en"
},
{
"name": "size_categories",
"value": "100B<n<1T"
},
{
"name": "annotations_creators",
"value": "no-annotation"
},
{
"name": "language_creators",
"value": "found"
},
{
"name": "pretty_name",
"value": "the Pile"
},
{
"name": "source_datasets",
"value": "original"
},
{
"name": "paperswithcode_id",
"value": "the-pile"
},
{
"name": "license",
"value": "other"
}
]
},
"description": "The Pile is a 825 GiB diverse, open source language modelling data set that consists of 22 smaller, high-quality\ndatasets combined together.",
"governance": {
"owners": [
{
"organization": {
"name": "EleutherAI",
"url": "https://huggingface.co/EleutherAI"
}
}
]
}
}
]
}
]
}