# rai by Prayas Abhinav: Open-Weight Model Source: https://savrn.com/models/rai Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve rai (33M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them. | Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits | | --- | --- | --- | --- | --- | --- | | 16-bit | 0.1 GB | 0.1 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 8-bit | 0.0 GB | 0.0 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 4-bit | 0.0 GB | 0.0 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the [SAVRN Index](https://savrn.com/ai-index/pricing/gpus), read Oct 9, 2026. [rai on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/rai/gpus) ## Model Card By Prayas Abhinav, published under cc-by-4.0, revision 0dbf31fd2c41. ### rai reader 0.3.7 — model card What it is. An extractive question-answering model. Given a question and a short typed answer, it returns the phrase in the answer that answers the question, or nothing. It cannot generate text. It is the reading part of [rai](https://github.com/koherarchitecture/rai), a tool that says only whether a description is complete under a notion of completeness written down in advance. Every decision after the reading is made in plain code. Trained from. [deepset/minilm-uncased-squad2](https://savrn.com/models/minilm-uncased-squad2) (CC-BY-4.0), deepset's fine-tune of Microsoft's [MiniLM-L12-H384-uncased](https://huggingface.co/microsoft/MiniLM-L12-H384-uncased) (MIT) on SQuAD 2.0. Same architecture, 33M parameters, nothing added. Credit to deepset and Microsoft. [Read the full model card (992 words)](https://savrn.com/models/rai/card) ## Configuration Architecture BertForQuestionAnswering Context length (tokens) 512 Layers 12 Hidden size 384 Feed-forward size 1,536 Attention heads 12 Vocabulary size 30,522 Model type bert ## Identity and Version Repository prayasabhinav/rai Publisher Prayas Abhinav Task Question answering Modality Text Library transformers Parameters 33M parameters Languages en Revision 0dbf31fd2c4156c288d43f1952a7e7fba8ee2972 First published 2026-09-22 Last updated 2026-09-27 ## Files and Weights 7 files, 133.6 MB in total. The weights are 1 file totalling 132.9 MB in safetensors. Weights1 file · 132.9 MB Configuration1 file · 720 B Tokenizer2 files · 712.2 KB Documentation2 files · 6.9 KB Repository1 file · 1.5 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model.safetensors | Weights | 132.9 MB | c0aab2b6767d | | config.json | Configuration | 720 B | — | | LICENSE.md | Documentation | 838 B | — | | README.md | Documentation | 6.1 KB | — | | .gitattributes | Repository | 1.5 KB | — | | tokenizer.json | Tokenizer | 711.6 KB | — | | tokenizer_config.json | Tokenizer | 598 B | — | ## License and Download License cc-by-4.0 Access Open weights, no gate Download size 132.9 MB [Download from Prayas Abhinav](https://huggingface.co/prayasabhinav/rai) Released by Prayas Abhinav through its official repository on Hugging Face. [Read the license](https://creativecommons.org/licenses/by/4.0/). ## Built From - Derived from [deepset/minilm-uncased-squad2](https://savrn.com/models/minilm-uncased-squad2) ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 132.9 MB | | 16-bit | 0.1 GB | | 8-bit | 0.0 GB | | 4-bit | 0.0 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Questions About rai ### How much GPU memory does rai need? About 0.1 GB at 16-bit and 0 GB at 4-bit: the weights (33M parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run rai on? At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### Can I use rai commercially? Yes. rai is released under Creative Commons Attribution 4.0. CC BY 4.0 permits sharing and adapting the work, including commercially, provided the creator is credited and changes are indicated. ### What is rai's context length? 512 tokens, from the maximum position embeddings in its published configuration. ## Similar Models Model · Question answering ### [minilm-uncased-squad2](https://savrn.com/models/minilm-uncased-squad2) [Deepset](https://savrn.com/model-publishers/deepset) Haystack is an AI orchestration framework to build customizable, production-ready LLM applications. You can use this model in Haystack to do extractive question answering on documents. To load and run the model with Haystack: For a complete example with an extractive question answering pipeline that scales over many documents, check out the corresponding Haystack tutorial. Evaluated on the SQuAD 2.0 dev set with the official eval script. Timo Möller: timo.moeller@deepset.ai deepset is the company behind the production-ready open-source AI framework Haystack. We also have a Discord community open to everyone! Open weights cc-by-4.0 33M parameters 512 tokens transformers [View model](https://savrn.com/models/minilm-uncased-squad2) Model · Question answering ### [basic-chat-model](https://savrn.com/models/basic-chat-model) [Lenin Villanueva](https://savrn.com/model-publishers/leninangelov) This modelcard aims to be a base template for new models. It has been generated using this raw template. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019). Open weights apache-2.0 31M parameters [View model](https://savrn.com/models/basic-chat-model) Model · Question answering ### [bert-medium-squad2-distilled](https://savrn.com/models/bert-medium-squad2-distilled) [Deepset](https://savrn.com/model-publishers/deepset) Haystack version 1.x distillation feature was used for training. deepset/bert-large-uncased-whole-word-masking-squad2 was used as the teacher model. Haystack is an AI orchestration framework to build customizable, production-ready LLM applications. You can use this model in Haystack to do extractive question answering on documents. To load and run the model with Haystack: For a complete example with an extractive question answering pipeline that scales over many documents, check out the corresponding Haystack tutorial. - Timo Möller: timo.moeller [at] deepset.ai deepset is the company behind the production-ready open-source AI framework Haystack. We also have a Discord community open to… Open weights mit 41M parameters 512 tokens transformers [View model](https://savrn.com/models/bert-medium-squad2-distilled) Model · Question answering ### [mobilebert-uncased-squad-v2](https://savrn.com/models/mobilebert-uncased-squad-v2) [Qingqing Cao](https://savrn.com/model-publishers/csarron) MobileBERT is a thin version of BERTLARGE, while equipped with bottleneck structures and a carefully designed balance between self-attentions and feed-forward networks. This model was fine-tuned from the HuggingFace checkpoint google/mobilebert-uncased on SQuAD2.0. CPU: Intel(R) Core(TM) i7-6800K CPU @ 3.40GHz Memory: 32 GiB GPUs: 2 GeForce GTX 1070, each with 8GiB memory GPU driver: 418.87.01, CUDA: 10.1 It took about 3.5 hours to finish. Note that the above results didn't involve any hyperparameter search. Open weights mit 25M parameters 512 tokens transformers [View model](https://savrn.com/models/mobilebert-uncased-squad-v2) Model · Question answering ### [mobilebert-uncased-squad-v1](https://savrn.com/models/mobilebert-uncased-squad-v1) [Qingqing Cao](https://savrn.com/model-publishers/csarron) MobileBERT is a thin version of BERTLARGE, while equipped with bottleneck structures and a carefully designed balance between self-attentions and feed-forward networks. This model was fine-tuned from the HuggingFace checkpoint google/mobilebert-uncased on SQuAD1.1. CPU: Intel(R) Core(TM) i7-6800K CPU @ 3.40GHz Memory: 32 GiB GPUs: 2 GeForce GTX 1070, each with 8GiB memory GPU driver: 418.87.01, CUDA: 10.1 It took about 3 hours to finish. Note that the above results didn't involve any hyperparameter search. Open weights mit 25M parameters 512 tokens transformers [View model](https://savrn.com/models/mobilebert-uncased-squad-v1) Model · Question answering ### [koelectra-small-v2-distilled-korquad-384](https://savrn.com/models/koelectra-small-v2-distilled-korquad-384) [Jangwon Park](https://savrn.com/model-publishers/monologg) Open weights 14M parameters 512 tokens transformers [View model](https://savrn.com/models/koelectra-small-v2-distilled-korquad-384) ## Prayas Abhinav [All models and datasets](https://savrn.com/model-publishers/prayasabhinav) ## Versions - [0dbf31fd2c41](https://savrn.com/models/rai/versions/0dbf31fd2c41) · current 2026-09-27 ## Explore More - [All question answering models](https://savrn.com/models/tasks/question-answering) - [All models under cc-by-4.0](https://savrn.com/models/licenses/cc-by-4-0) - [Model comparisons](https://savrn.com/models/comparisons) - [The model directory](https://savrn.com/models) - [Open model prices by host](https://savrn.com/ai-index/pricing/open-models) ## Source - Repository metadata, read 2026-09-27. - [Hugging Face record](https://huggingface.co/prayasabhinav/rai) - [How the hub is built](https://savrn.com/model-hub/methodology)