Segatron-XL large, M=384 + HCP
No estimate
No hardware requirements for this model
The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.
On record
Full specification
Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.
Origin
Who built this model, where, and when it was published.
- Organisation
- Microsoft Research,University of Waterloo
- Organisation type
- Industry,Academia
- Country
- United States of America, Canada
- Published
- 21 March 2022
- Authors
- He Bai, Tong Wang, Alessandro Sordoni, Peng Shi
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Language
- Task
- Language modeling
Size
How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.
- Parameters
- 257M
- Training data
- 103,000,000 tokens
- Epochs
- 167.02
257M (Table 1) 18 layers, 16 heads, hidden size 1024 batch size 128.
103000000 tokens - size of wikitext 103 Training steps: 350,000 steps Sequence length: 384 tokens Batch size: 128 Tokens per batch = 384 tokens/batch × 128 = 49,152 tokens/batch Total tokens = 49,152 tokens/batch × 350,000 steps = 17.2 billion tokens 17200000000/103000000 = 167 epochs
Training compute
The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.
- Training compute
- 2.7 × 10¹⁹ FLOP
- How it was established
- Operation counting
6 FLOP / token / parameter * 257000000 parameters * 384 tokens/batch × 128 [batch size] * 350000 steps = 2.6527334e+19 FLOP
Availability
Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.
- Weights
- Closed — provider access only
- Model access
- Unreleased
- Training code
- Open (non-commercial)
code, no license https://github.com/richardbaihe/robustlm
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Why it is tracked
- SOTA improvement
- Record confidence
- Confident
- Citations
- 17
- Benchmark data
- Segatron-XL large, M=384 + HCP
"Empirically, this curriculum learning strategy consistently improves perplexity over various large, highly-performant state-of-the-art Transformer-based models on two datasets, WikiText-103 and ARXIV"
Sources
Where this record came from and when it was last checked.
- Reference
- Better Language Model with Hypernym Class Prediction
- Last updated
- 25 May 2026
What the numbers mean
Where it came from
Segatron-XL large, M=384 + HCP was published by Microsoft Research,University of Waterloo, in United States of America, in March 2022. It comes out of industry,Academia.
It works in Language, and is recorded as doing language modeling.
Because the weights are not available, none of the hardware figures elsewhere on this site apply to it.
What went into building it
Producing it required around 2.7 × 10¹⁹ FLOP of arithmetic, which is a statement about the training budget rather than about inference.
The training set ran to roughly 103,000,000 tokens.
Its inclusion criterion is sOTA improvement.
Answers
Segatron-XL large, M=384 + HCP — common questions
Who created Segatron-XL large, M=384 + HCP?
Segatron-XL large, M=384 + HCP was published by Microsoft Research,University of Waterloo, based in United States of America, categorised as industry,Academia.
When was Segatron-XL large, M=384 + HCP released?
Segatron-XL large, M=384 + HCP was published in March 2022. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.
What is Segatron-XL large, M=384 + HCP used for?
Segatron-XL large, M=384 + HCP works in Language, and is recorded as handling language modeling. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.
How much compute was used to train Segatron-XL large, M=384 + HCP?
Around 2.7 × 10¹⁹ FLOP. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.
What GPU do I need to run Segatron-XL large, M=384 + HCP?
None. Segatron-XL large, M=384 + HCP is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.
Is Segatron-XL large, M=384 + HCP open source?
No. Segatron-XL large, M=384 + HCP has not had its weights published, so it exists only as a service controlled by its owner.
How many parameters does Segatron-XL large, M=384 + HCP have?
Segatron-XL large, M=384 + HCP has 257M parameters. 257M (Table 1) 18 layers, 16 heads, hidden size 1024 batch size 128. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.
The other direction
Looking at it from the other side?
This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.