selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

InternScience logo

Can the RTX A6000 run MLExpert?

MLExpert·34.7B (3.5B active)·RTX A6000·48 GB VRAM

Saved in this browser, used on every model page.
Yes

Yes. The RTX A6000 runs MLExpert fully in VRAM at Q8_0 (about 37 GB of 48 GB), at roughly 115 tokens/s.

best quant
Q8_0
needs
37 GB
usable
48 GB
est. speed
~115 tok/s

Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the MLExpert page →

Every quantisation of MLExpert on the RTX A6000

QuantQualityMemoryFitSpeed
Q2_K Heavy loss15 GBfits~372 tok/s
Q3_K_M Noticeable loss18 GBfits~250 tok/s
Q4_K_M Small loss23 GBfits~202 tok/s
Q5_K_M Near-lossless28 GBfits~172 tok/s
Q6_K Near-lossless31 GBfits~148 tok/s
Q8_0 pickLossless37 GBfits~115 tok/s
FP16 / BF16 Lossless69 GBoffload~13 tok/s

Speed is a bandwidth-based estimate for generating text (768 GB/s); prompt processing and long contexts are slower.

Measured speed on the RTX A6000

No community reports for this pair yet.

Report your own speed →

Run MLExpert on the RTX A6000

Using Q8_0, the best fit for RTX A6000.

Ollama · Q8_0 straight from Hugging Facedocs
ollama run hf.co/bartowski/InternScience_MLExpert-GGUF:Q8_0
llama.cpp · OpenAI-compatible server with a web UIdocs
llama-server -hf bartowski/InternScience_MLExpert-GGUF:Q8_0

Downloads the GGUF on first run. Installed through llama.app? The same flags work with `llama serve`.

LM Studio · Download with the lms CLIdocs
lms get bartowski/InternScience_MLExpert-GGUF@Q8_0

Then load it from the LM Studio app or with `lms load`.

vLLM · Serve an OpenAI-compatible APIdocs
vllm serve InternScience/MLExpert
vLLM · Docker Compose (NVIDIA)docs
services:
  vllm:
    image: vllm/vllm-openai:latest
    command: InternScience/MLExpert
    ports:
      - "8000:8000"
    ipc: host
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

API on port 8000.

SGLang · Serve an OpenAI-compatible APIdocs
sglang serve InternScience/MLExpert --host 0.0.0.0 --port 30000
SGLang · Docker Compose (NVIDIA)docs
services:
  sglang:
    image: lmsysorg/sglang:latest
    command: sglang serve InternScience/MLExpert --host 0.0.0.0 --port 30000
    ports:
      - "30000:30000"
    ipc: host
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

API on port 30000.

Transformers · Chat with the pipeline APIdocs
from transformers import pipeline

pipe = pipeline("text-generation", model="InternScience/MLExpert", dtype="auto", device_map="auto")
messages = [{"role": "user", "content": "Explain self-hosting in one paragraph."}]
out = pipe(messages, max_new_tokens=512)
print(out[0]["generated_text"][-1]["content"])

Chat models that fit the RTX A6000 comfortably

All that fit →
ModelParamsBest quantMemorySpeedLicence
36BA3BQ8_0≈38 GBFits in memory~133 tok/sCommercial OK
31.2BA4.6BQ8_0≈39 GBFits in memory~86 tok/sCommercial OK
32.8BQ8_0≈37 GBFits in memory~12 tok/sCommercial OK
RecursiveSI logoflintRecursiveSI
39.8BA1.3BQ8_0≈43 GBFits in memory~318 tok/sCommercial OK
20.9BA4.2BQ8_0≈13 GBFits in memory~95 tok/sCommercial OK
30.5BA3BQ8_0≈33 GBFits in memory~133 tok/sCommercial OK

Other GPUs that run MLExpert

Frequently asked questions

Can the RTX A6000 run MLExpert?
Yes. The RTX A6000 runs MLExpert fully in VRAM at Q8_0 (about 37 GB of 48 GB), at roughly 115 tokens/s.
How fast does MLExpert run on the RTX A6000?
An estimate from memory bandwidth gives about 115 tokens/s at Q8_0.
Which quantisation of MLExpert should I use on the RTX A6000?
Q8_0: the largest one that fits. Expected quality: lossless compared with the full-precision weights.

Memory and speed are estimates from the model's size and the card's memory bandwidth; real numbers depend on the runtime, drivers and context length. Community figures are medians of user reports.