Ovis-Omni-Embedding-3B-gp…mixed-w4-w4-w8 by Hung Cheung Chan

Runs On

What it takes to serve Ovis-Omni-Embedding-3B-gptq-mixed-w4-w4-w8 (2,935 parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.

Precision	Cheapest setup	Per hour	Also fits
16-bit	1x MI300X (192 GB) Vultr	$1.85	1x H100 $1.99 · 1x MI325X $2.00
8-bit	1x MI300X (192 GB) Vultr	$1.85	1x H100 $1.99 · 1x MI325X $2.00
4-bit	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 Oct 9, 2026.

Ovis-Omni-Embedding-3B-gptq-mixed-w4-w4-w8 on every accelerator the SAVRN Index prices, at every precision

Model Card

Experimental calibration-based weight-only GPTQ variant of The Thinker transformer uses 4-bit weights for layers 0–26 and 8-bit weights for layer 27. Audio, vision, talker, token2wav, embeddings, and norms remain in the original precision. Calibration used eight short text samples with LLM Compressor 0.14.0 and compressed-tensors 0.19.0. This checkpoint includes a packaging repair: the compressor export contained invalid group scales, so scales were recomputed from the original BF16 weights per group before the vLLM test. Treat this as an experimental GPTQ-derived checkpoint and benchmark retrieval quality before production use. Tested with vLLM 0.30.0 on an 8-GiB RTX 3080 Laptop GPU: The…

Excerpt from the card by Hung Cheung Chan.

Configuration

Architecture: Qwen2_5OmniForConditionalGeneration
Hidden size: 2,048
Model type: qwen2_5_omni
Quantization: compressed-tensors

Identity and Version

Repository: pt810/Ovis-Omni-Embedding-3B-gptq-mixed-w4-w4-w8
Publisher: Hung Cheung Chan
Task: Text to audio
Modality: Other
Library: transformers
Parameters: 2,935 parameters
Languages: Not stated by the source
Revision: 667b6a7a0058b0c652342618fd7f18111448beb1
First published: 2026-09-24
Last updated: 2026-09-24

Files and Weights

257 files, 8.0 GB in total. The weights are 232 files totalling 7.9 GB in bin, pt, pth, safetensors.

Weights232 files · 7.9 GB

Configuration8 files · 438.6 KB

Tokenizer2 files · 11.4 MB

Documentation2 files · 13.5 KB

Other12 files · 69.9 MB

Repository1 file · 1.9 KB

Every file

File	Type	Size	SHA-256
gptq-noscale-part-00001.safetensors	Weights	622.3 MB	18e3b3c10b04
gptq-noscale-part-00002.safetensors	Weights	22.5 MB	9faaa35b1719
gptq-noscale-part-00003.safetensors	Weights	27.3 MB	914234460360
gptq-noscale-part-00004.safetensors	Weights	27.3 MB	9da1c6b28176
gptq-noscale-part-00005.safetensors	Weights	22.5 MB	71f61323f5b4
gptq-noscale-part-00006.safetensors	Weights	27.3 MB	281e0b97df4b
gptq-noscale-part-00007.safetensors	Weights	27.3 MB	f4fdbbac908e
gptq-noscale-part-00008.safetensors	Weights	22.5 MB	4d483b3eb7f2
gptq-noscale-part-00009.safetensors	Weights	27.3 MB	a79ab4a4f2a9
gptq-noscale-part-00010.safetensors	Weights	27.3 MB	597761192896
gptq-noscale-part-00011.safetensors	Weights	22.5 MB	26b7bf622cda
gptq-noscale-part-00012.safetensors	Weights	27.3 MB	ac183259a828
gptq-noscale-part-00013.safetensors	Weights	27.3 MB	861ace164f58
gptq-noscale-part-00014.safetensors	Weights	22.5 MB	f604893088b6
gptq-noscale-part-00015.safetensors	Weights	27.3 MB	facf92bb8a53
gptq-noscale-part-00016.safetensors	Weights	27.3 MB	b7705719dad4
gptq-noscale-part-00017.safetensors	Weights	22.5 MB	bc1c70f6e6c4
gptq-noscale-part-00018.safetensors	Weights	27.3 MB	56bd379167f3
gptq-noscale-part-00019.safetensors	Weights	27.3 MB	872ab109fb5a
gptq-noscale-part-00020.safetensors	Weights	22.5 MB	e2be4acebe6c
gptq-noscale-part-00021.safetensors	Weights	27.3 MB	9492e34cbe18
gptq-noscale-part-00022.safetensors	Weights	27.3 MB	1cb36de02be2
gptq-noscale-part-00023.safetensors	Weights	22.5 MB	7f50fd3a242f
gptq-noscale-part-00024.safetensors	Weights	27.3 MB	437d622906cc
gptq-noscale-part-00025.safetensors	Weights	27.3 MB	e74b34c3a2f8
gptq-noscale-part-00026.safetensors	Weights	22.5 MB	b472844d2e97
gptq-noscale-part-00027.safetensors	Weights	27.3 MB	e567059d1c70
gptq-noscale-part-00028.safetensors	Weights	27.3 MB	e470e66c4e29
gptq-noscale-part-00029.safetensors	Weights	22.5 MB	fe43932a9f62
gptq-noscale-part-00030.safetensors	Weights	27.3 MB	99d7969ae15a
gptq-noscale-part-00031.safetensors	Weights	27.3 MB	6a211813a6ba
gptq-noscale-part-00032.safetensors	Weights	22.5 MB	eef8c672ebfb
gptq-noscale-part-00033.safetensors	Weights	22.5 MB	23a2f6880d63
gptq-noscale-part-00034.safetensors	Weights	32.0 MB	7d4ea732b63b
gptq-noscale-part-00035.safetensors	Weights	45.1 MB	603b164c37d8
gptq-noscale-part-00036.safetensors	Weights	45.1 MB	715082875d7a
gptq-noscale-part-00037.safetensors	Weights	45.1 MB	e5bae4c8db49
gptq-noscale-part-00038.safetensors	Weights	18.9 MB	9f5943516290
gptq-noscale-part-00039.safetensors	Weights	45.1 MB	43afa938a2cd
gptq-noscale-part-00040.safetensors	Weights	45.1 MB	c988aad20794
gptq-noscale-part-00041.safetensors	Weights	45.1 MB	ab6edf352b8d
gptq-noscale-part-00042.safetensors	Weights	30.2 MB	70eef0ab56c6
gptq-noscale-part-00043.safetensors	Weights	27.3 MB	5fe72d8b2cdf
gptq-noscale-part-00044.safetensors	Weights	45.1 MB	4c50be882345
gptq-noscale-part-00045.safetensors	Weights	45.1 MB	f6047f068350
gptq-noscale-part-00046.safetensors	Weights	45.1 MB	9317316a0fe6
gptq-noscale-part-00047.safetensors	Weights	18.9 MB	fba5ee18dade
gptq-noscale-part-00048.safetensors	Weights	45.1 MB	dce8e4815b92
gptq-noscale-part-00049.safetensors	Weights	45.1 MB	7fb6ba530825
gptq-noscale-part-00050.safetensors	Weights	45.1 MB	53d6ef525a32
gptq-noscale-part-00051.safetensors	Weights	18.9 MB	91b51e4f9993
gptq-noscale-part-00052.safetensors	Weights	45.1 MB	7ede65d676cc
gptq-noscale-part-00053.safetensors	Weights	45.1 MB	e2fa62a78c59
gptq-noscale-part-00054.safetensors	Weights	45.1 MB	cd93aff371bd
gptq-noscale-part-00055.safetensors	Weights	18.9 MB	e3aa0622f278
gptq-noscale-part-00056.safetensors	Weights	45.1 MB	9662916121dd
gptq-noscale-part-00057.safetensors	Weights	45.1 MB	e06657b37aa5
gptq-noscale-part-00058.safetensors	Weights	45.1 MB	210b7701cdc6
gptq-noscale-part-00059.safetensors	Weights	18.9 MB	cf4bce48db61
gptq-noscale-part-00060.safetensors	Weights	45.1 MB	14915382f2b2
gptq-noscale-part-00061.safetensors	Weights	45.1 MB	d5db4e6eabb8
gptq-noscale-part-00062.safetensors	Weights	45.1 MB	15b0f23e4b44
gptq-noscale-part-00063.safetensors	Weights	18.9 MB	53999c46b372
gptq-noscale-part-00064.safetensors	Weights	45.1 MB	ea534282439c
gptq-noscale-part-00065.safetensors	Weights	45.1 MB	21dd12bd0154
gptq-noscale-part-00066.safetensors	Weights	45.1 MB	9247dc62ccff
gptq-noscale-part-00067.safetensors	Weights	30.2 MB	86712ba7267a
gptq-noscale-part-00068.safetensors	Weights	27.3 MB	4e9e2548a137
gptq-noscale-part-00069.safetensors	Weights	22.5 MB	e7eb172fcf40
gptq-noscale-part-00070.safetensors	Weights	27.3 MB	912e87ce96d6
gptq-noscale-part-00071.safetensors	Weights	27.3 MB	3fb1b988030f
gptq-noscale-part-00072.safetensors	Weights	22.5 MB	7a7484abd8f0
gptq-noscale-part-00073.safetensors	Weights	27.3 MB	7bcde5fb5d84
gptq-noscale-part-00074.safetensors	Weights	27.3 MB	fc2b8f7868c7
gptq-noscale-part-00075.safetensors	Weights	22.5 MB	dc54627530aa
gptq-noscale-part-00076.safetensors	Weights	16.0 MB	aa0dca939a13
gptq-repaired-scales.safetensors	Weights	33.7 MB	e549f8216c8a
orig-1-part-00001.safetensors	Weights	15.1 MB	97ea623c6e64
orig-1-part-00002.safetensors	Weights	34.6 MB	ee2c9cca465a
orig-1-part-00003.safetensors	Weights	29.8 MB	ffba8993f637
orig-1-part-00004.safetensors	Weights	29.8 MB	556d18d09c70
orig-1-part-00005.safetensors	Weights	29.8 MB	a0e93080691b
orig-1-part-00006.safetensors	Weights	29.8 MB	94ab920503ac
orig-1-part-00007.safetensors	Weights	29.8 MB	3de6c27c9c2c
orig-1-part-00008.safetensors	Weights	29.8 MB	772e48eacbc8
orig-1-part-00009.safetensors	Weights	29.8 MB	68a35817b81a
orig-1-part-00010.safetensors	Weights	29.8 MB	27a9089d4266
orig-1-part-00011.safetensors	Weights	29.8 MB	64a681dfa415
orig-1-part-00012.safetensors	Weights	29.8 MB	35ae3cd3494a
orig-1-part-00013.safetensors	Weights	29.8 MB	f7b9dda8d51b
orig-1-part-00014.safetensors	Weights	29.8 MB	1839d1af699d
orig-1-part-00015.safetensors	Weights	29.8 MB	09d8d44ecce8
orig-1-part-00016.safetensors	Weights	29.8 MB	4a293b505b21
orig-1-part-00017.safetensors	Weights	29.8 MB	491918d98c0a
orig-1-part-00018.safetensors	Weights	29.8 MB	7e7a7942f83a
orig-1-part-00019.safetensors	Weights	29.8 MB	d53449b60343
orig-1-part-00020.safetensors	Weights	29.8 MB	c7fdd7509c00
orig-1-part-00021.safetensors	Weights	29.8 MB	349e771520b3
orig-1-part-00022.safetensors	Weights	29.8 MB	d473493531a6
orig-1-part-00023.safetensors	Weights	29.8 MB	4ebd559dc7fc
orig-1-part-00024.safetensors	Weights	29.8 MB	316f253530cf
orig-1-part-00025.safetensors	Weights	29.8 MB	e0967bb66672
orig-1-part-00026.safetensors	Weights	33.5 MB	e046a3d2c3ec
orig-1-part-00027.safetensors	Weights	23.9 MB	194510a670f9
orig-1-part-00028.safetensors	Weights	26.2 MB	0c2834b5ac28
orig-1-part-00029.safetensors	Weights	32.8 MB	f87e968c5c26
orig-1-part-00030.safetensors	Weights	32.8 MB	5dc7f9aa72c2
orig-1-part-00031.safetensors	Weights	26.2 MB	e5ad85262944
orig-1-part-00032.safetensors	Weights	26.2 MB	6e6821df278c
orig-1-part-00033.safetensors	Weights	32.8 MB	62c7def02ed4
orig-1-part-00034.safetensors	Weights	32.8 MB	c82fc91b9d4e
orig-1-part-00035.safetensors	Weights	26.2 MB	c36fd5dc2337
orig-1-part-00036.safetensors	Weights	26.2 MB	ba9d699e88d4
orig-1-part-00037.safetensors	Weights	32.8 MB	586fcee4e85a
orig-1-part-00038.safetensors	Weights	32.8 MB	bd6226e448fc
orig-1-part-00039.safetensors	Weights	26.2 MB	c47e1612743b
orig-1-part-00040.safetensors	Weights	26.2 MB	ad48b7c17e6d
orig-1-part-00041.safetensors	Weights	32.8 MB	436ffe6c0c09
orig-1-part-00042.safetensors	Weights	32.8 MB	359e6dec9d7a
orig-1-part-00043.safetensors	Weights	26.2 MB	74f1f73d4518
orig-1-part-00044.safetensors	Weights	26.2 MB	015e83f081be
orig-1-part-00045.safetensors	Weights	32.8 MB	d3dc5ebf8d42
orig-1-part-00046.safetensors	Weights	32.8 MB	79546275c76e
orig-1-part-00047.safetensors	Weights	26.2 MB	c79d4e3b8273
orig-1-part-00048.safetensors	Weights	26.2 MB	255c317e34e5
orig-1-part-00049.safetensors	Weights	32.8 MB	0252d67663dd
orig-1-part-00050.safetensors	Weights	32.8 MB	0918927446c9
orig-1-part-00051.safetensors	Weights	26.2 MB	0dd3166ba927
orig-1-part-00052.safetensors	Weights	26.2 MB	1726bb174e8f
orig-1-part-00053.safetensors	Weights	32.8 MB	1cdad91c71b8
orig-1-part-00054.safetensors	Weights	32.8 MB	d409411f3b17
orig-1-part-00055.safetensors	Weights	26.2 MB	01541461aeff
orig-1-part-00056.safetensors	Weights	26.2 MB	72c04ae152fe
orig-1-part-00057.safetensors	Weights	32.8 MB	fe4de7256e58
orig-1-part-00058.safetensors	Weights	32.8 MB	d17f79d91600
orig-1-part-00059.safetensors	Weights	26.2 MB	8449892eac68
orig-1-part-00060.safetensors	Weights	26.2 MB	82e0205ea49d
orig-1-part-00061.safetensors	Weights	32.8 MB	2608100ef2dd
orig-1-part-00062.safetensors	Weights	32.8 MB	28896af2b1a8
orig-1-part-00063.safetensors	Weights	26.2 MB	2b6f7a5a4aab
orig-1-part-00064.safetensors	Weights	26.2 MB	20323b60898d
orig-1-part-00065.safetensors	Weights	32.8 MB	bde71feb5d24
orig-1-part-00066.safetensors	Weights	32.8 MB	2191032350a0
orig-1-part-00067.safetensors	Weights	26.2 MB	56c68760873b
orig-1-part-00068.safetensors	Weights	26.2 MB	0f950f4c8fae
orig-1-part-00069.safetensors	Weights	32.8 MB	b7918c01d6a2
orig-1-part-00070.safetensors	Weights	11.8 MB	853f44dad306
orig-1-part-00071.safetensors	Weights	622.3 MB	053dddc78ee2
orig-2-part-00001.safetensors	Weights	30.6 MB	3f840b599146
orig-2-part-00002.safetensors	Weights	30.6 MB	70151986fc71
orig-2-part-00003.safetensors	Weights	30.6 MB	989a4ce1cfd5
orig-2-part-00004.safetensors	Weights	32.8 MB	4a85ba273866
orig-2-part-00005.safetensors	Weights	32.8 MB	ddf259cea1c7
orig-2-part-00006.safetensors	Weights	30.6 MB	6f5d98b417cc
orig-2-part-00007.safetensors	Weights	30.6 MB	8dba9579e591
orig-2-part-00008.safetensors	Weights	30.6 MB	d1214df2d5eb
orig-2-part-00009.safetensors	Weights	32.8 MB	7dbb55b374a4
orig-2-part-00010.safetensors	Weights	32.8 MB	280e2c6631a4
orig-2-part-00011.safetensors	Weights	30.6 MB	c3784d0eee0a
orig-2-part-00012.safetensors	Weights	30.6 MB	a81c4ed68e07
orig-2-part-00013.safetensors	Weights	30.6 MB	c13139dabae5
orig-2-part-00014.safetensors	Weights	32.8 MB	dea03e69cc56
orig-2-part-00015.safetensors	Weights	32.8 MB	55599c12ddbe
orig-2-part-00016.safetensors	Weights	30.6 MB	5c836cc4cb58
orig-2-part-00017.safetensors	Weights	30.6 MB	89da203dd46f
orig-2-part-00018.safetensors	Weights	30.6 MB	d3a09325762c
orig-2-part-00019.safetensors	Weights	32.8 MB	9809b881a39f
orig-2-part-00020.safetensors	Weights	32.8 MB	9137a4eed957
orig-2-part-00021.safetensors	Weights	30.6 MB	f5998be5287f
orig-2-part-00022.safetensors	Weights	30.6 MB	0da834fd84b7
orig-2-part-00023.safetensors	Weights	30.6 MB	89b597352fd7
orig-2-part-00024.safetensors	Weights	32.8 MB	a1c21c9dbfed
orig-2-part-00025.safetensors	Weights	32.8 MB	e0bc7c0c35b5
orig-2-part-00026.safetensors	Weights	30.6 MB	f41d32226684
orig-2-part-00027.safetensors	Weights	30.6 MB	6421ec668f77
orig-2-part-00028.safetensors	Weights	30.6 MB	2fa8a94017a6
orig-2-part-00029.safetensors	Weights	32.8 MB	a99e833eaa06
orig-2-part-00030.safetensors	Weights	27.4 MB	8f73d67bb4e7
orig-2-part-00031.safetensors	Weights	30.6 MB	08d9861ac06e
orig-2-part-00032.safetensors	Weights	30.6 MB	a240351082be
orig-2-part-00033.safetensors	Weights	32.8 MB	56e04178fd7c
orig-2-part-00034.safetensors	Weights	32.8 MB	a0a8bc8aae2b
orig-2-part-00035.safetensors	Weights	30.6 MB	f73475797e70
orig-2-part-00036.safetensors	Weights	30.6 MB	dd1fca0cba13
orig-2-part-00037.safetensors	Weights	17.5 MB	c934947d4cc6
orig-3-part-00001.safetensors	Weights	27.4 MB	dfa66d4dce3a
orig-3-part-00002.safetensors	Weights	30.6 MB	ff9d5c90cbcc
orig-3-part-00003.safetensors	Weights	30.6 MB	302c6fd43307
orig-3-part-00004.safetensors	Weights	26.3 MB	98f44d141639
orig-3-part-00005.safetensors	Weights	52.4 MB	ff8a9b594ecb
orig-3-part-00006.safetensors	Weights	32.8 MB	0aaa54ff2d8d
orig-3-part-00007.safetensors	Weights	30.7 MB	9dc454320c80
orig-3-part-00008.safetensors	Weights	33.6 MB	bf67a98d500b
orig-3-part-00009.safetensors	Weights	28.2 MB	13896baf33f4
orig-3-part-00010.safetensors	Weights	26.0 MB	63b7db35cd82
orig-3-part-00011.safetensors	Weights	33.4 MB	aab31ba06b9a
orig-3-part-00012.safetensors	Weights	33.2 MB	9c8ef6dbe454
orig-3-part-00013.safetensors	Weights	32.2 MB	1c1924147f9e
orig-3-part-00014.safetensors	Weights	31.9 MB	ee1b3a0031d2
orig-3-part-00015.safetensors	Weights	31.5 MB	3028472d32f6
orig-3-part-00016.safetensors	Weights	29.4 MB	ae660d6e3c3c
orig-3-part-00017.safetensors	Weights	29.4 MB	44db5d504dde
orig-3-part-00018.safetensors	Weights	29.4 MB	b97bed2cdd84
orig-3-part-00019.safetensors	Weights	29.4 MB	dc4a04925edb
orig-3-part-00020.safetensors	Weights	29.4 MB	273523e5523c
orig-3-part-00021.safetensors	Weights	29.4 MB	cbd3c0cc08ca
orig-3-part-00022.safetensors	Weights	29.4 MB	4828bcb45ed3
orig-3-part-00023.safetensors	Weights	29.4 MB	72e0584e7a2f
orig-3-part-00024.safetensors	Weights	29.4 MB	6ad465c5d7e4
orig-3-part-00025.safetensors	Weights	29.4 MB	265b3d244dda
orig-3-part-00026.safetensors	Weights	29.4 MB	c5ce6debe6c6
orig-3-part-00027.safetensors	Weights	29.4 MB	3be6359a4d21
orig-3-part-00028.safetensors	Weights	29.4 MB	83d1b0c23cbf
orig-3-part-00029.safetensors	Weights	29.4 MB	00e95b6e0655
orig-3-part-00030.safetensors	Weights	29.4 MB	f4c7dfb193a7
orig-3-part-00031.safetensors	Weights	29.4 MB	7fe5d86688e0
orig-3-part-00032.safetensors	Weights	29.4 MB	67c993c6db63
orig-3-part-00033.safetensors	Weights	29.4 MB	48bf5444f80e
orig-3-part-00034.safetensors	Weights	29.4 MB	5e04579b8051
orig-3-part-00035.safetensors	Weights	29.4 MB	194e4ade4393
orig-3-part-00036.safetensors	Weights	21.0 MB	f00f30fc829b
rng_state_0.pth	Weights	16.4 KB	478b41e9f26d
rng_state_1.pth	Weights	16.4 KB	ce29a8767a7d
rng_state_2.pth	Weights	16.4 KB	61a48db01164
rng_state_3.pth	Weights	16.4 KB	b9562ee82247
rng_state_4.pth	Weights	16.4 KB	e7d2767d83c3
rng_state_5.pth	Weights	16.4 KB	76816358d4e5
rng_state_6.pth	Weights	16.4 KB	1562e7520c97
rng_state_7.pth	Weights	16.4 KB	a6b6cabaed04
scheduler.pt	Weights	1.5 KB	2947d2764fc5
spk_dict.pt	Weights	259.5 KB	6a05609b28f5
training_args.bin	Weights	8.7 KB	efe2b1b63155
args.json	Configuration	96 B	—
config.json	Configuration	16.0 KB	—
generation_config.json	Configuration	154 B	—
model.safetensors.index.json	Configuration	278.9 KB	—
preprocessor_config.json	Configuration	667 B	—
processor_config.json	Configuration	2.8 KB	—
trainer_state.json	Configuration	106.8 KB	—
zero_to_fp32.py	Configuration	33.3 KB	—
LICENSE	Documentation	11.5 KB	—
README.md	Documentation	2.0 KB	—
chat_template.jinja	Other	1.3 KB	—
latest	Other	15 B	—
model-00001-of-00003.safetensors.download.log	Other	73.5 KB	—
model-00002-of-00003.safetensors.download.log	Other	72.9 KB	—
model-00003-of-00003.safetensors.download.log	Other	73.4 KB	—
ovis_blog_table1.pdf	Other	484.4 KB	ae5707e6bc8e
ovis_blog_table1.png	Other	436.4 KB	e6cdde834fdb
ovis_embedding_data_centric.png	Other	729.9 KB	0ad28dd7a765
ovis_embedding_model_architecture.png	Other	401.5 KB	da256408d299
ovis_embedding_train_inference.png	Other	414.1 KB	49353acc8881
ovis_logo.png	Other	85.5 KB	—
test.part	Other	67.1 MB	f4c42971269d
.gitattributes	Repository	1.9 KB	—
tokenizer.json	Tokenizer	11.4 MB	1ab7a851e5c6
tokenizer_config.json	Tokenizer	939 B	—

License and Download

License: Not stated by the source
Access: Open weights, no gate
Download size: 7.9 GB

Download from Hung Cheung Chan

Released by Hung Cheung Chan through its official repository on Hugging Face.

Built From

Derived from ATH-MaaS/Ovis-Omni-Embedding-3B
Quantized from ATH-MaaS/Ovis-Omni-Embedding-3B

Memory Requirements

Precision	Weights in memory
As published	7.9 GB
16-bit	0.0 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 Ovis-Omni-Embedding-3B-gptq-mixed-w4-w4-w8

How much GPU memory does Ovis-Omni-Embedding-3B-gptq-mixed-w4-w4-w8 need?

About 0 GB at 16-bit and 0 GB at 4-bit: the weights (2,935 parameters) plus a working margin. A long context needs more.

What is the cheapest GPU to run Ovis-Omni-Embedding-3B-gptq-mixed-w4-w4-w8 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.

Similar Models

Model · Text to audio

musicgen-medium

AI at Meta

MusicGen is a text-to-music model capable of genreating high-quality music samples conditioned on text descriptions or audio prompts. It is a single stage auto-regressive Transformer model trained over a 32kHz EnCodec tokenizer with 4 codebooks sampled at 50 Hz. Unlike existing methods, like MusicLM, MusicGen doesn't require a self-supervised semantic representation, and it generates all 4 codebooks in one pass. By introducing a small delay between the codebooks, we show we can predict them in parallel, thus having only 50 auto-regressive steps per second of audio. MusicGen was published in Simple and Controllable Music Generation by Jade Copet, Felix Kreuk, Itai Gat, Tal Remez, David Kant…

Open weights cc-by-nc-4.0 transformers

View model

Model · Text to audio

MiniMax-Music3-GGUF

Audio.cpp

GGUF package for MiniMax Music 3 for audio.cpp. Star our repo so you don't miss important updates! https://github.com/0xShug0/audio.cpp Upstream license: https://huggingface.co/MiniMaxAI/MiniMax-Music3/blob/main/LICENSE - The implementation is available on the main branch and release 0.6.1. - The current runtime uses model-local resource loading instead of treating the v1 spec as the runtime contract. This keeps component selection flexible while the package layout and option surface settle. - The default component mix favors Q40 for the large language model and flow transformer, with Q80 for the RVQ depth decoder. - BF16, Q80, and Q40 component variants are included for quality/performance…

Open weights other audio.cpp

View model

Model · Text to audio

Yue2-3B-GGUF

Audio.cpp

This repository contains audio.cpp-native GGUF weights for Yue2-3B. The model has been merged into the main branch. LoRA support added in release 0.8.1. The demos below were generated with yue2-3b-bf16.gguf and yue2-vae-f32.gguf. The q8.wav comparison files were generated with yue2-3b-q80.gguf and yue2-vae-f16.gguf using the same prompts and seeds. The q40.wav comparison files were generated with yue2-3b-q40.gguf and yue2-vae-f16.gguf. Replace with your audio.cpp checkout and with this downloaded GGUF repo directory. To generate the Q40 versions, use the same commands and replace --session-option yue2.modelgguf=yue2-3b-bf16.gguf with --session-option yue2.modelgguf=yue2-3b-q40.gguf, and use…

Open weights cc-by-nc-4.0 audio.cpp

View model

Model · Text to audio

yue2-mltnt-militant-reggae

Because Reasons

Four artist-style LoRAs that push YuE2-3B into modern militant roots reggae: dark raspy male patois vocals, steppers and one-drop grooves, deep sub bass, bubbling Hammond, nyabinghi drums, horn stabs, dub sirens and spring reverb. Conscious, apocalyptic, anthemic. Each file patches both halves of YuE2 in one go: the autoregressive planner (writes the score, decides the arrangement and the vocal lines) and the flow-matching decoder (the sound). Trigger word for all three: mltnt. All demos use the same original lyric, seed 7, 32 steps dpm2 / sgmuniform, no post-processing. MLTNT Frontline — baseline recipe, prompt prompts/steppersbaseline.txt, dense lyric (verses written at ~17 words per…

Open weights cc-by-nc-4.0

View model

Model · Text to audio

igbo-mms-tts

Omeziri Zion

This is the model card of a transformers model that has been pushed on the Hub. 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 36M parameters transformers

View model

Model · Text to audio

akan-twi-mms

Abdul Rashid Dickson

This is the model card of a transformers model that has been pushed on the Hub. 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 cc-by-nc-4.0 83M parameters transformers

View model