InferX Catalog | translategemma-27b-it-FP8-Dynamic

translategemma-27b-it-FP8-Dynamic

Multilingual translation model optimized with FP8 for fast, memory-efficient inference

kaitchup text text2text low-latency

Log in to deploy

kaitchup/translategemma-27b-it-FP8-Dynamic is a 27B instruction-tuned translation model based on Gemma, optimized with FP8 dynamic quantization to reduce memory usage and improve throughput while maintaining high-quality multilingual translation, making it ideal for production translation services and latency-sensitive deployments.

Metadata

Provider

kaitchup

Modality

text

API type

text2text

Source

huggingface / kaitchup/translategemma-27b-it-FP8-Dynamic

Created

2026-03-31 23:37:25 UTC

Updated

2026-03-31 23:56:57 UTC

Catalog version

1

Visibility

Published

Specifications

Parameters

27.00B

MoE

No

Max model length

2048

Image

vllm/vllm-openai:v0.9.0

Default Deploy Config

GPU count

1

vRAM

32000 MB

Summary

1xGPU 32000 MB

Recommended Use Cases

Translation

Model Spec

{
    "image": "vllm/vllm-openai:v0.9.0",
    "commands": [
        "--model",
        "/root/.cache/huggingface/hub/models--kaitchup--translategemma-27b-it-FP8-Dynamic/snapshots/6bdba8b2d7d690be008ce87348f2ce4be2e8a876",
        "--disable-custom-all-reduce",
        "--trust-remote-code",
        "--gpu-memory-utilization",
        "0.95",
        "--max-model-len",
        "2048",
        "--chat-template",
        "/root/.cache/huggingface/hub/models--kaitchup--translategemma-27b-it-FP8-Dynamic/snapshots/6bdba8b2d7d690be008ce87348f2ce4be2e8a876/chat_template.jinja",
        "--enable-prefix-caching",
        "--prefix-caching-hash-algo",
        "builtin"
    ],
    "resources": {
        "GPU": {
            "Count": 1,
            "vRam": 32000
        }
    },
    "envs": [],
    "policy": {
        "Obj": {
            "min_replica": 0,
            "max_replica": 1,
            "standby_per_node": 1,
            "parallel": 50,
            "queue_len": 100,
            "queue_timeout": 30.0,
            "scalein_timeout": 1.0,
            "scaleout_policy": {
                "WaitQueueRatio": {
                    "wait_ratio": 0.1
                }
            },
            "runtime_config": {
                "graph_sync": false
            }
        }
    },
    "sample_query": {
        "body": {
            "stream": "true",
            "max_tokens": "1000",
            "temperature": "0"
        },
        "path": "v1/completions",
        "prompt": "write a quick sort algorithm.",
        "apiType": "text2text",
        "dataUrl": "",
        "prompts": [],
        "loadingTimeout": 90
    }
}