Model reference · open weights
Lumina-mGPT-768 is an open-weight language model from Alpha-VLLM. Lumina-mGPT-7B-768 (BF16) weighs 14.0 GB; the smallest configuration that runs it is 2× RTX 3060 12 GB.
What it is
| Released by | Alpha-VLLM |
|---|---|
| Released | 2024-07-29 |
| Parameters | 7.0B |
| VRAM | 14.0 GB for the weights |
What it runs on
| Card | Requests at once | Context max | Memory |
|---|---|---|---|
| 4K each | |||
| RTX 3060 12 GB | — | — | 11.6 GB |
| RTX 4060 Ti 16 GB | — | 1K | 15.4 GB |
| RTX 3090 24 GB | 4 | all 4K | 23.4 GB |
| RTX 4090 24 GB | 4 | all 4K | 23.4 GB |
| RTX 5090 32 GB | 7 | all 4K | 31.0 GB |
| L40S 48 GB | 13 | all 4K | 44.0 GB |
| A100 80 GB | 29 | all 4K | 78.2 GB |
| H100 80 GB | 27 | all 4K | 78.1 GB |
| RTX PRO 6000 Blackwell 96 GB | 35 | all 4K | 93.8 GB |
| DGX Spark (GB10) 128 GB unified | 41 | all 4K | 107 GB |
| H200 141 GB | 55 | all 4K | 138 GB |
| B200 180 GB | 73 | all 4K | 176 GB |
| 2× RTX 3060 12 GB tensor parallel | 3 | all 4K | 11.6 GB a card |
| 2× RTX 4060 Ti 16 GB tensor parallel | 7 | all 4K | 15.4 GB a card |
| 2× RTX 4090 24 GB tensor parallel | 14 | all 4K | 23.4 GB a card |
| 2× RTX 3090 24 GB tensor parallel | 14 | all 4K | 23.4 GB a card |
| 2× RTX 5090 32 GB tensor parallel | 21 | all 4K | 31.0 GB a card |
| Requests at once | 4K tokens each |
|---|---|
| 1 | 16.8 GB |
| 5 | 25.4 GB |
| 8 | 31.9 GB |
| 16 | 49.0 GB |
| 32 | 83.4 GB |
| 64 | 152 GB |
One card, with vLLM's small-card settings.
From the model card
Lumina-mGPT is a family of multimodal autoregressive models capable of various vision and language tasks, particularly excelling in generating flexible photorealistic images from text descriptions.
We provide the implementation of Lumina-mGPT, as well as sampling code, in our github repository.
Quoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.