SAVRN
Search Contact SAVRN

Open-weight model

b-58m

by Isai Lab aixk/b-58m

Parameters58M
Context512
Weights1.5 GB
License
AccessOpen weights
Monthly Downloads

Runs On

What it takes to serve b-58m (58M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.

PrecisionWeightsMemory neededCheapest setupPer hourAlso fits
16-bit 0.1 GB 0.1 GB 1x MI300X (192 GB)
Vultr
$1.85 1x H100 $1.99 · 1x MI325X $2.00
8-bit 0.1 GB 0.1 GB 1x MI300X (192 GB)
Vultr
$1.85 1x H100 $1.99 · 1x MI325X $2.00
4-bit 0.0 GB 0.0 GB 1x MI300X (192 GB)
Vultr
$1.85 1x H100 $1.99 · 1x 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, read Sep 18, 2026.

Model Card

The publisher has not written a card for this model.

Configuration

Architecture
FHN_T4Max_150M
Context length (tokens)
512
Layers
12
Hidden size
768
Vocabulary size
1,536
Stored precision
float16
Model type
fhn_concept_gau

Identity and Version

Repository
aixk/b-58m
Publisher
Isai Lab
Task
Not stated by the source
Modality
Other
Library
Not stated by the source
Parameters
58M parameters
Languages
Not stated by the source
Revision
301ceaa9ab8400cd5df40291777985936d593781
First published
2026-09-16
Last updated
2026-09-18

Files and Weights

21 files, 1.8 GB in total. The weights are 3 files totalling 1.5 GB in pt, safetensors.

Weights3 files · 1.5 GB
Configuration2 files · 1.2 KB
Tokenizer1 file · 275.2 KB
Other14 files · 339.7 MB
Repository1 file · 2.3 KB
Every file
FileTypeSizeSHA-256
checkpoint_step_128000.ptWeights689.2 MB cb426264f80d
checkpoint_step_129000.ptWeights689.2 MB 0e301cbb1f78
model.safetensorsWeights116.4 MB 65b4adbe76ac
config.jsonConfiguration350 B
vlite_manifest.jsonConfiguration896 B
model.vliteOther28.5 MB 2e4b976a9c46
model_full_int1.vliteOther14.7 MB f2fd3e428ca3
model_full_int2.vliteOther21.7 MB 45681a789f12
model_full_int3.vliteOther28.6 MB 3b68d5bf0fa9
model_full_int4.vliteOther35.5 MB daa900775670
model_micro_int1.vliteOther25.8 MB 95e1f9e9d90f
model_micro_int2.vliteOther28.5 MB 2e4b976a9c46
model_micro_int3.vliteOther31.0 MB 53f5d5dc0ac8
model_micro_int4.vliteOther33.1 MB 7a33511d8e70
model_shared.vliteOther24.0 MB 8df3b0b7778b
model_shared_int1.vliteOther10.1 MB bd841d4dd668
model_shared_int2.vliteOther14.7 MB ccee5cd516b3
model_shared_int3.vliteOther19.4 MB 11292fa9596e
model_shared_int4.vliteOther24.0 MB 8df3b0b7778b
.gitattributesRepository2.3 KB
tokenizer_state.jsonTokenizer275.2 KB

License and Download

License
Not stated by the source
Access
Open weights, no gate
Download size
1.5 GB
Download from Isai Lab

Released by Isai Lab through its official repository on Hugging Face.

Memory Requirements

PrecisionWeights in memory
As published1.5 GB
16-bit0.1 GB
8-bit0.1 GB
4-bit0.0 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.

Questions About b-58m

How much GPU memory does b-58m need?

About 0.1 GB at 16-bit and 0 GB at 4-bit: the weights (58M parameters) plus a working margin. A long context needs more.

What is the cheapest GPU to run b-58m 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.

What is b-58m's context length?

512 tokens, from the maximum position embeddings in its published configuration.