bytedance / sa2va-8b-image
Sa2VA: Marrying SAM2 with LLaVA for Dense Grounded Understanding of Images and Videos
Prediction
bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aaID8bma77cdyxrm80cn5g5tecbpcmStatusSucceededSourceWebHardwareL40STotal durationCreatedInput
{ "image": "https://replicate.delivery/pbxt/MXdtMQfOr4E3fyfRIXQneZUR7TTUWWMRhje7CV1hhqvsREkv/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", "instruction": "what is the giraffe doing?" }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", { input: { image: "https://replicate.delivery/pbxt/MXdtMQfOr4E3fyfRIXQneZUR7TTUWWMRhje7CV1hhqvsREkv/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", instruction: "what is the giraffe doing?" } } ); console.log(output);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", input={ "image": "https://replicate.delivery/pbxt/MXdtMQfOr4E3fyfRIXQneZUR7TTUWWMRhje7CV1hhqvsREkv/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", "instruction": "what is the giraffe doing?" } ) print(output)
To learn more, take a look at the guide on getting started with Python.
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", "input": { "image": "https://replicate.delivery/pbxt/MXdtMQfOr4E3fyfRIXQneZUR7TTUWWMRhje7CV1hhqvsREkv/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", "instruction": "what is the giraffe doing?" } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa \ -i 'image="https://replicate.delivery/pbxt/MXdtMQfOr4E3fyfRIXQneZUR7TTUWWMRhje7CV1hhqvsREkv/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp"' \ -i 'instruction="what is the giraffe doing?"'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MXdtMQfOr4E3fyfRIXQneZUR7TTUWWMRhje7CV1hhqvsREkv/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", "instruction": "what is the giraffe doing?" } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
The giraffe is eating leaves from a tree.<|im_end|>{ "completed_at": "2025-02-22T05:05:10.386006Z", "created_at": "2025-02-22T05:05:09.623000Z", "data_removed": false, "error": null, "id": "8bma77cdyxrm80cn5g5tecbpcm", "input": { "image": "https://replicate.delivery/pbxt/MXdtMQfOr4E3fyfRIXQneZUR7TTUWWMRhje7CV1hhqvsREkv/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", "instruction": "what is the giraffe doing?" }, "logs": null, "metrics": { "predict_time": 0.755694414, "total_time": 0.763006 }, "output": { "img": null, "response": "The giraffe is eating leaves from a tree.<|im_end|>" }, "started_at": "2025-02-22T05:05:09.630311Z", "status": "succeeded", "urls": { "get": "https://api.replicate.com/v1/predictions/8bma77cdyxrm80cn5g5tecbpcm", "cancel": "https://api.replicate.com/v1/predictions/8bma77cdyxrm80cn5g5tecbpcm/cancel" }, "version": "956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa" }
Prediction
bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aaInput
{ "image": "https://replicate.delivery/pbxt/MXdtc5yJDPoUGs6li6sYevHiNXWJjaD9O4kvCwYIAIWTHWsG/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", "instruction": "segment the giraffe" }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", { input: { image: "https://replicate.delivery/pbxt/MXdtc5yJDPoUGs6li6sYevHiNXWJjaD9O4kvCwYIAIWTHWsG/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", instruction: "segment the giraffe" } } ); console.log(output);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", input={ "image": "https://replicate.delivery/pbxt/MXdtc5yJDPoUGs6li6sYevHiNXWJjaD9O4kvCwYIAIWTHWsG/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", "instruction": "segment the giraffe" } ) print(output)
To learn more, take a look at the guide on getting started with Python.
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", "input": { "image": "https://replicate.delivery/pbxt/MXdtc5yJDPoUGs6li6sYevHiNXWJjaD9O4kvCwYIAIWTHWsG/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", "instruction": "segment the giraffe" } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa \ -i 'image="https://replicate.delivery/pbxt/MXdtc5yJDPoUGs6li6sYevHiNXWJjaD9O4kvCwYIAIWTHWsG/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp"' \ -i 'instruction="segment the giraffe"'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MXdtc5yJDPoUGs6li6sYevHiNXWJjaD9O4kvCwYIAIWTHWsG/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", "instruction": "segment the giraffe" } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-02-22T05:05:26.289057Z", "created_at": "2025-02-22T05:05:25.409000Z", "data_removed": false, "error": null, "id": "rd2vw06bm5rma0cn5g5rmesp78", "input": { "image": "https://replicate.delivery/pbxt/MXdtc5yJDPoUGs6li6sYevHiNXWJjaD9O4kvCwYIAIWTHWsG/replicate-prediction-1spvj2jc8hrm80cn5f6t1xxg4m.webp", "instruction": "segment the giraffe" }, "logs": "propagate in video: 0%| | 0/1 [00:00<?, ?it/s]\npropagate in video: 100%|██████████| 1/1 [00:00<00:00, 5599.87it/s]", "metrics": { "predict_time": 0.872783281, "total_time": 0.880057 }, "output": { "img": "https://replicate.delivery/xezq/94rYVsLtK0IfLiKPefc5zxeeXhFcGf0acrzUNY6uVSfVLa4IKA/output.png", "response": "Sure, [SEG] .<|im_end|>" }, "started_at": "2025-02-22T05:05:25.416274Z", "status": "succeeded", "urls": { "get": "https://api.replicate.com/v1/predictions/rd2vw06bm5rma0cn5g5rmesp78", "cancel": "https://api.replicate.com/v1/predictions/rd2vw06bm5rma0cn5g5rmesp78/cancel" }, "version": "956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa" }
Prediction
bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aaInput
{ "image": "https://replicate.delivery/pbxt/MXdtxHiZZNR2n01EApC0novzn2N3DvtSV1sVydnPfdsbLUA7/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", "instruction": "caption this" }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", { input: { image: "https://replicate.delivery/pbxt/MXdtxHiZZNR2n01EApC0novzn2N3DvtSV1sVydnPfdsbLUA7/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", instruction: "caption this" } } ); console.log(output);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", input={ "image": "https://replicate.delivery/pbxt/MXdtxHiZZNR2n01EApC0novzn2N3DvtSV1sVydnPfdsbLUA7/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", "instruction": "caption this" } ) print(output)
To learn more, take a look at the guide on getting started with Python.
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", "input": { "image": "https://replicate.delivery/pbxt/MXdtxHiZZNR2n01EApC0novzn2N3DvtSV1sVydnPfdsbLUA7/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", "instruction": "caption this" } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa \ -i 'image="https://replicate.delivery/pbxt/MXdtxHiZZNR2n01EApC0novzn2N3DvtSV1sVydnPfdsbLUA7/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp"' \ -i 'instruction="caption this"'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MXdtxHiZZNR2n01EApC0novzn2N3DvtSV1sVydnPfdsbLUA7/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", "instruction": "caption this" } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
A snowboarder in a white jacket and red helmet is riding down a snowy slope, skillfully navigating the terrain.<|im_end|>{ "completed_at": "2025-02-22T05:05:48.467758Z", "created_at": "2025-02-22T05:05:47.344000Z", "data_removed": false, "error": null, "id": "d6jbr991a1rma0cn5g6av83wvg", "input": { "image": "https://replicate.delivery/pbxt/MXdtxHiZZNR2n01EApC0novzn2N3DvtSV1sVydnPfdsbLUA7/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", "instruction": "caption this" }, "logs": null, "metrics": { "predict_time": 1.115205571, "total_time": 1.123758 }, "output": { "img": null, "response": "A snowboarder in a white jacket and red helmet is riding down a snowy slope, skillfully navigating the terrain.<|im_end|>" }, "started_at": "2025-02-22T05:05:47.352553Z", "status": "succeeded", "urls": { "get": "https://api.replicate.com/v1/predictions/d6jbr991a1rma0cn5g6av83wvg", "cancel": "https://api.replicate.com/v1/predictions/d6jbr991a1rma0cn5g6av83wvg/cancel" }, "version": "956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa" }
Prediction
bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aaID4zhmqqvdkxrm80cn5g6aqpxcrmStatusSucceededSourceWebHardwareL40STotal durationCreatedInput
{ "image": "https://replicate.delivery/pbxt/MXduFcHTkeo9XAeatyWaSHr97nm2uNQMopHdM5GnSEZUXZcd/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", "instruction": "segment the snowboarder" }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", { input: { image: "https://replicate.delivery/pbxt/MXduFcHTkeo9XAeatyWaSHr97nm2uNQMopHdM5GnSEZUXZcd/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", instruction: "segment the snowboarder" } } ); console.log(output);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", input={ "image": "https://replicate.delivery/pbxt/MXduFcHTkeo9XAeatyWaSHr97nm2uNQMopHdM5GnSEZUXZcd/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", "instruction": "segment the snowboarder" } ) print(output)
To learn more, take a look at the guide on getting started with Python.
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", "input": { "image": "https://replicate.delivery/pbxt/MXduFcHTkeo9XAeatyWaSHr97nm2uNQMopHdM5GnSEZUXZcd/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", "instruction": "segment the snowboarder" } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa \ -i 'image="https://replicate.delivery/pbxt/MXduFcHTkeo9XAeatyWaSHr97nm2uNQMopHdM5GnSEZUXZcd/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp"' \ -i 'instruction="segment the snowboarder"'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MXduFcHTkeo9XAeatyWaSHr97nm2uNQMopHdM5GnSEZUXZcd/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", "instruction": "segment the snowboarder" } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-02-22T05:06:07.535785Z", "created_at": "2025-02-22T05:06:06.879000Z", "data_removed": false, "error": null, "id": "4zhmqqvdkxrm80cn5g6aqpxcrm", "input": { "image": "https://replicate.delivery/pbxt/MXduFcHTkeo9XAeatyWaSHr97nm2uNQMopHdM5GnSEZUXZcd/replicate-prediction-x9kew5a699rm80cn584apv9w7w.webp", "instruction": "segment the snowboarder" }, "logs": "propagate in video: 0%| | 0/1 [00:00<?, ?it/s]\npropagate in video: 100%|██████████| 1/1 [00:00<00:00, 6345.39it/s]", "metrics": { "predict_time": 0.650710074, "total_time": 0.656785 }, "output": { "img": "https://replicate.delivery/xezq/fQxAun650enr0UcAsFPRhrBnVRhH0MHVI960o4V473FfphjoA/output.png", "response": "Sure, [SEG] .<|im_end|>" }, "started_at": "2025-02-22T05:06:06.885075Z", "status": "succeeded", "urls": { "get": "https://api.replicate.com/v1/predictions/4zhmqqvdkxrm80cn5g6aqpxcrm", "cancel": "https://api.replicate.com/v1/predictions/4zhmqqvdkxrm80cn5g6aqpxcrm/cancel" }, "version": "956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa" }
Prediction
bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aaID10md0xdcjhrma0cn5g68mqfhx0StatusSucceededSourceWebHardwareL40STotal durationCreatedInput
{ "image": "https://replicate.delivery/pbxt/MXduVZDRkqG8GUs1taAsUbXBjMyqBkiy96KueCYbrqTeGBjL/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", "instruction": "segment the dinosaur to the left of the sunglasses" }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", { input: { image: "https://replicate.delivery/pbxt/MXduVZDRkqG8GUs1taAsUbXBjMyqBkiy96KueCYbrqTeGBjL/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", instruction: "segment the dinosaur to the left of the sunglasses" } } ); console.log(output);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", input={ "image": "https://replicate.delivery/pbxt/MXduVZDRkqG8GUs1taAsUbXBjMyqBkiy96KueCYbrqTeGBjL/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", "instruction": "segment the dinosaur to the left of the sunglasses" } ) print(output)
To learn more, take a look at the guide on getting started with Python.
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", "input": { "image": "https://replicate.delivery/pbxt/MXduVZDRkqG8GUs1taAsUbXBjMyqBkiy96KueCYbrqTeGBjL/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", "instruction": "segment the dinosaur to the left of the sunglasses" } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa \ -i 'image="https://replicate.delivery/pbxt/MXduVZDRkqG8GUs1taAsUbXBjMyqBkiy96KueCYbrqTeGBjL/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp"' \ -i 'instruction="segment the dinosaur to the left of the sunglasses"'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MXduVZDRkqG8GUs1taAsUbXBjMyqBkiy96KueCYbrqTeGBjL/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", "instruction": "segment the dinosaur to the left of the sunglasses" } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-02-22T05:06:23.688469Z", "created_at": "2025-02-22T05:06:22.996000Z", "data_removed": false, "error": null, "id": "10md0xdcjhrma0cn5g68mqfhx0", "input": { "image": "https://replicate.delivery/pbxt/MXduVZDRkqG8GUs1taAsUbXBjMyqBkiy96KueCYbrqTeGBjL/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", "instruction": "segment the dinosaur to the left of the sunglasses" }, "logs": "propagate in video: 0%| | 0/1 [00:00<?, ?it/s]\npropagate in video: 100%|██████████| 1/1 [00:00<00:00, 4821.04it/s]", "metrics": { "predict_time": 0.684259284, "total_time": 0.692469 }, "output": { "img": "https://replicate.delivery/xezq/I9Ho1jDOE5ZeEyjF5lVom4h7tulMG8MrS3Ic4SeydL7P1wRUA/output.png", "response": "Sure, [SEG] .<|im_end|>" }, "started_at": "2025-02-22T05:06:23.004209Z", "status": "succeeded", "urls": { "get": "https://api.replicate.com/v1/predictions/10md0xdcjhrma0cn5g68mqfhx0", "cancel": "https://api.replicate.com/v1/predictions/10md0xdcjhrma0cn5g68mqfhx0/cancel" }, "version": "956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa" }
Prediction
bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aaID8n9kqjpj1xrma0cn5g6bvs479mStatusSucceededSourceWebHardwareL40STotal durationCreatedInput
{ "image": "https://replicate.delivery/pbxt/MXduf6AK8CULojAal2w7NzxMAuxePl44vAKJCleSPOqJeCOu/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", "instruction": "how many feathers are there?" }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", { input: { image: "https://replicate.delivery/pbxt/MXduf6AK8CULojAal2w7NzxMAuxePl44vAKJCleSPOqJeCOu/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", instruction: "how many feathers are there?" } } ); console.log(output);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", input={ "image": "https://replicate.delivery/pbxt/MXduf6AK8CULojAal2w7NzxMAuxePl44vAKJCleSPOqJeCOu/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", "instruction": "how many feathers are there?" } ) print(output)
To learn more, take a look at the guide on getting started with Python.
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", "input": { "image": "https://replicate.delivery/pbxt/MXduf6AK8CULojAal2w7NzxMAuxePl44vAKJCleSPOqJeCOu/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", "instruction": "how many feathers are there?" } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa \ -i 'image="https://replicate.delivery/pbxt/MXduf6AK8CULojAal2w7NzxMAuxePl44vAKJCleSPOqJeCOu/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp"' \ -i 'instruction="how many feathers are there?"'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MXduf6AK8CULojAal2w7NzxMAuxePl44vAKJCleSPOqJeCOu/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", "instruction": "how many feathers are there?" } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
There are two feathers in the image.<|im_end|>{ "completed_at": "2025-02-22T05:06:33.283870Z", "created_at": "2025-02-22T05:06:32.591000Z", "data_removed": false, "error": null, "id": "8n9kqjpj1xrma0cn5g6bvs479m", "input": { "image": "https://replicate.delivery/pbxt/MXduf6AK8CULojAal2w7NzxMAuxePl44vAKJCleSPOqJeCOu/replicate-prediction-t89m1g7t1drma0cn551bhpaep0.webp", "instruction": "how many feathers are there?" }, "logs": null, "metrics": { "predict_time": 0.685370321, "total_time": 0.69287 }, "output": { "img": null, "response": "There are two feathers in the image.<|im_end|>" }, "started_at": "2025-02-22T05:06:32.598500Z", "status": "succeeded", "urls": { "get": "https://api.replicate.com/v1/predictions/8n9kqjpj1xrma0cn5g6bvs479m", "cancel": "https://api.replicate.com/v1/predictions/8n9kqjpj1xrma0cn5g6bvs479m/cancel" }, "version": "956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa" }
Prediction
bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aaIDq14aa81ek5rme0cn5g6vvp5h4gStatusSucceededSourceWebHardwareL40STotal durationCreatedInput
{ "image": "https://replicate.delivery/pbxt/MXdv23uHMMctP6BlTJHJA8qiufRVjVH8uW2ujUK5yFLk2smC/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", "instruction": "how many ducks are there?" }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", { input: { image: "https://replicate.delivery/pbxt/MXdv23uHMMctP6BlTJHJA8qiufRVjVH8uW2ujUK5yFLk2smC/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", instruction: "how many ducks are there?" } } ); console.log(output);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", input={ "image": "https://replicate.delivery/pbxt/MXdv23uHMMctP6BlTJHJA8qiufRVjVH8uW2ujUK5yFLk2smC/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", "instruction": "how many ducks are there?" } ) print(output)
To learn more, take a look at the guide on getting started with Python.
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", "input": { "image": "https://replicate.delivery/pbxt/MXdv23uHMMctP6BlTJHJA8qiufRVjVH8uW2ujUK5yFLk2smC/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", "instruction": "how many ducks are there?" } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa \ -i 'image="https://replicate.delivery/pbxt/MXdv23uHMMctP6BlTJHJA8qiufRVjVH8uW2ujUK5yFLk2smC/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp"' \ -i 'instruction="how many ducks are there?"'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MXdv23uHMMctP6BlTJHJA8qiufRVjVH8uW2ujUK5yFLk2smC/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", "instruction": "how many ducks are there?" } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
There is one yellow rubber duck in the drawer.<|im_end|>{ "completed_at": "2025-02-22T05:06:57.034258Z", "created_at": "2025-02-22T05:06:56.281000Z", "data_removed": false, "error": null, "id": "q14aa81ek5rme0cn5g6vvp5h4g", "input": { "image": "https://replicate.delivery/pbxt/MXdv23uHMMctP6BlTJHJA8qiufRVjVH8uW2ujUK5yFLk2smC/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", "instruction": "how many ducks are there?" }, "logs": null, "metrics": { "predict_time": 0.746140453, "total_time": 0.753258 }, "output": { "img": null, "response": "There is one yellow rubber duck in the drawer.<|im_end|>" }, "started_at": "2025-02-22T05:06:56.288117Z", "status": "succeeded", "urls": { "get": "https://api.replicate.com/v1/predictions/q14aa81ek5rme0cn5g6vvp5h4g", "cancel": "https://api.replicate.com/v1/predictions/q14aa81ek5rme0cn5g6vvp5h4g/cancel" }, "version": "956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa" }
Prediction
bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aaID0aft2enjkhrme0cn5g6rtbma2gStatusSucceededSourceWebHardwareL40STotal durationCreatedInput
{ "image": "https://replicate.delivery/pbxt/MXdvYnwacQqM9mJxYgq568FOlFGwpod7gDuEZXfXgkKQ5C53/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", "instruction": "please segment the yellow duck" }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", { input: { image: "https://replicate.delivery/pbxt/MXdvYnwacQqM9mJxYgq568FOlFGwpod7gDuEZXfXgkKQ5C53/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", instruction: "please segment the yellow duck" } } ); console.log(output);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", input={ "image": "https://replicate.delivery/pbxt/MXdvYnwacQqM9mJxYgq568FOlFGwpod7gDuEZXfXgkKQ5C53/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", "instruction": "please segment the yellow duck" } ) print(output)
To learn more, take a look at the guide on getting started with Python.
Run bytedance/sa2va-8b-image using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "bytedance/sa2va-8b-image:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa", "input": { "image": "https://replicate.delivery/pbxt/MXdvYnwacQqM9mJxYgq568FOlFGwpod7gDuEZXfXgkKQ5C53/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", "instruction": "please segment the yellow duck" } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa \ -i 'image="https://replicate.delivery/pbxt/MXdvYnwacQqM9mJxYgq568FOlFGwpod7gDuEZXfXgkKQ5C53/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp"' \ -i 'instruction="please segment the yellow duck"'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/bytedance/sa2va-8b-image@sha256:956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MXdvYnwacQqM9mJxYgq568FOlFGwpod7gDuEZXfXgkKQ5C53/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", "instruction": "please segment the yellow duck" } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-02-22T05:07:30.746641Z", "created_at": "2025-02-22T05:07:30.076000Z", "data_removed": false, "error": null, "id": "0aft2enjkhrme0cn5g6rtbma2g", "input": { "image": "https://replicate.delivery/pbxt/MXdvYnwacQqM9mJxYgq568FOlFGwpod7gDuEZXfXgkKQ5C53/replicate-prediction-eey39qeqh1rme0cn552tca7k0c.webp", "instruction": "please segment the yellow duck" }, "logs": "propagate in video: 0%| | 0/1 [00:00<?, ?it/s]\npropagate in video: 100%|██████████| 1/1 [00:00<00:00, 6462.72it/s]", "metrics": { "predict_time": 0.661875876, "total_time": 0.670641 }, "output": { "img": "https://replicate.delivery/xezq/acmv8OUvToIwJROKsbjSwb39eWHFtDeGSQH3eVIRkORlshjoA/output.png", "response": "Sure, [SEG] .<|im_end|>" }, "started_at": "2025-02-22T05:07:30.084766Z", "status": "succeeded", "urls": { "get": "https://api.replicate.com/v1/predictions/0aft2enjkhrme0cn5g6rtbma2g", "cancel": "https://api.replicate.com/v1/predictions/0aft2enjkhrme0cn5g6rtbma2g/cancel" }, "version": "956baf05a8a81ab47f1d0dac8eab6585b899790f342975a964840c4e9c63c7aa" }
Want to make some of these yourself?
Run this model