InternLM2 models
5 InternLM2 models on Hugging Face, from 7.7B to 19.9B parameters. At the precision each one is published in, the smallest needs about 17.3 GB of VRAM (internlm2-chat-7b-ExPO, cheapest live fit: RTX A5000) and the largest about 44.4 GB (internlm2-chat-20b-ExPO, cheapest live fit: RTX A6000). The cheapest way to run internlm2-chat-7b-ExPO is $0.176/hr.
InternLM2 models
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| internlm2-chat-7b-ExPO | 7.7B | BF16 | 17.3 GB | RTX A5000 | $0.176/hr |
| internlm2_5-7b-chat | 7.7B | BF16 | 17.3 GB | RTX A5000 | $0.176/hr |
| internlm2-chat-7b | 7.7B | BF16 | 17.3 GB | RTX A5000 | $0.176/hr |
| internlm2-chat-20b-ExPO | 19.9B | BF16 | 44.4 GB | RTX A6000 | $0.363/hr |
| internlm2-chat-20b | 19.9B | BF16 | 44.4 GB | RTX A6000 | $0.363/hr |
VRAM is for the precision the model is published in, with the same overhead and an 8,192-token context assumed on every page; see the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.
