Text generation · served with vLLM · ↓ 73.2K downloads a month · View on Hugging Face
farbodtavakkoli/OTel-2.0-LLM-31B-IT-QLoRA needs about 77.3 GB of GPU memory for conversations of up to 8,192 tokens (an estimate: its settings are behind its licence). It runs on 2x RTX A6000 from $0.98/h right now; best value: RTX PRO 6000 at $1.26/h.
The model's publisher asks users to accept its licence (apache-2.0). GridPup cannot download this one yet: choose an ungated version below.
Weights
63 GB
GPU memory needed
about 77.3 GB (estimate)
Conversation length
8,192 tokens
Memory · disk
39.3 GB · 109 GB
Architecture
Gemma4ForConditionalGeneration
Licence
apache-2.0
Hardware and deploy
The publisher has not let GridPup download this licence-gated model yet. An ungated version of it (above) can be deployed.
Questions
How much GPU memory does OTel-2.0-LLM-31B-IT-QLoRA need?
About 77.3 GB for conversations of up to 8,192 tokens: 62.55 GB of weights, the cache for one full-length conversation and the runtime's own memory (an estimate: its settings are behind its licence).
What is the cheapest GPU to run OTel-2.0-LLM-31B-IT-QLoRA?
Right now 2x RTX A6000 at $0.98 an hour; RTX PRO 6000 at $1.26 an hour is the best value for its speed. Prices are live offers from GPU clouds, checked every 15 minutes.
Can OTel-2.0-LLM-31B-IT-QLoRA run on an RTX 4090?
Not on one: it needs about 77.3 GB and an RTX 4090 has 24 GB. Use a bigger GPU, or a smaller (quantized) version of the model.
The memory is estimated from the weights: GridPup cannot read this model's settings until its licence is accepted.
Prices are per hour for the whole machine and its disk, right now. Speed is Vast's DLPerf score of the machine, relative to the cheapest. Best value: the most speed per dollar. Only machines with a Vast reliability above 95% that download at 400 Mbit/s or more are offered: the machine fetches the model itself.