Image-Text-to-Text
Transformers
Safetensors
English
Chinese
glm
glm-5.3-flash
dgx-spark
serving-recipe
tensorfold
speculative-decoding
dflash
shapleymcg
Instructions to use jakejharris/jspark3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use jakejharris/jspark3 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-text-to-text", model="jakejharris/jspark3")# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("jakejharris/jspark3", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use jakejharris/jspark3 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "jakejharris/jspark3" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "jakejharris/jspark3", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/jakejharris/jspark3
- SGLang
How to use jakejharris/jspark3 with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "jakejharris/jspark3" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "jakejharris/jspark3", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "jakejharris/jspark3" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "jakejharris/jspark3", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use jakejharris/jspark3 with Docker Model Runner:
docker model run hf.co/jakejharris/jspark3
Commit History
27 days ago
-
Restore pre-patch card content with version and link updates
Verified signaturea265fd6 -
Remove patch acceptance copy from hero
Verified signatured3074c4 -
Clarify known issues continue to apply to JSpark3 v2.0.2
Verified signature138aad3 -
Update JSpark3 card for v2.0.2 image fixes
Verified signature5294559
28 days ago
-
Card: link v2.0.1 errata and maintained install guides (jspark3 PR #25)
Verified signaturef3bbf4d -
Card: point the 1.5 s headline at known issue 17
Verified signature0be8ff3 -
docs: correct mikolaj92 credit; known issues 2 and 17
Verified signatureCard source: github.com/jakejharris/jspark3 PR #23, main 4f275648. README.md only.7ed6a13 -
Move the v1.8.x EXL3 weight and config files into v1.8.x-exl3/
Verified signatureServer-side copy of the v1.8.x EXL3/TR3 files from the repository root into v1.8.x-exl3/, with the root copies removed, in one commit. LFS files are copied by sha256, so no weight was re-uploaded. The v1.8.4 rollback keeps working from the pinned revision e6cb0b09b3bf9f2ce35721426c570c8e714c5fc4, which still serves the old root paths. README.md now names the folder.9b51dbb -
docs: point the v1.8.4 rollback at the pinned revision e6cb0b09
Verified signatureMirror of jakejharris/jspark3 GitHub PR #21 (merge 4df7344f). v1.8.4 guide downloads from Mia-AiLab, no longer public; the card now gives the replacement download pinned to e6cb0b09, which carries v1.8.4 SHA256SUMS cb0da1f9.6392ea7 -
Card metadata: drop the exl3 tag, add dflash and zh
Verified signaturev2.0.1 does not use the EXL3 files this repository keeps for the v1.8.4 rollback. Mirrors jakejharris/jspark3 PR #20 (merge ac66c1a6).b595809 -
docs: complete v2.0.1 credits and clarify the RigMark lead
Verified signature086ccb6
29 days ago
-
JSpark3 v2.0.1 card: lead with RigMark
Verified signature1df8909 -
JSpark3 v2.0.1 card
Verified signature0270d91
-
docs(card): credit Brandon M. Music for the checkpoint
Verified signature4b1629b -
JSpark3 v1.8.4 qualified coop release card
Verified signatureb706cec
-
docs(card): credit Brandon M. Music as quant author
Verified signature40a62a9 -
JSpark3 v1.8.3 release pointers
Verified signature483c5a4 -
docs: point the JSpark3 card at v1.8.2
Verified signature2076a4d
-
Correct model card license metadata to ShapleyMcg License v1.0
Verified signaturec7bbfc8 -
JSpark3 v1.8.0 recipe and results
Verified signature2d2ba24
-
Fix comparison table fit and give Recipe column the remaining width
Verified signature4be3dce
-
docs: refresh model card and add JSpark3 mark
Verified signatureDocumentation only. No weights, config, tokenizer, license, or mirror-receipt changes.1cf9f93 -
Update README.md
Verified signature04b8b9d
-
docs: publish JSpark3 v1.1.0 Cadence card and release guide
Verified signatureDocumentation only. No weights, config, tokenizer, license, or mirror-receipt changes.8e379cf
-
docs: publish measured headline results
jakejharrise25648f -
docs: lead model card with results and recipe
jakejharrise6cb0b0 -
docs: clarify weight provenance and license boundaries
Verified signatureClarify that the mirrored weights are Brandon M. Music's exact EXL3/TR3 quantization, re-hosted by Mia-AiLab and unchanged by JSpark3; distinguish the Apache-2.0 recipe from the ShapleyMcg-licensed weights and separate DFlash2 dependency.e986f41 -
Finalize public mirror status after verified merge
Verified signaturee7c34db -
Add the verified JSpark3 mirrored checkpoint (#1)
- Add files using upload-large-folder tool (79254e6a935200d717a81569dc7b61a95136b43f) - Add files using upload-large-folder tool (5798ed5acdde9592683c8616ffbc57413b6b2c64) - Add files using upload-large-folder tool (2bd4fa166bff0ddc82ce282453cc9f414e91bc02) - Record verified mirror completion (a85e4b12051893ac263a7f2b49c1d19a31e259ef) - Sync final v1.0.0 release metadata before mirror merge (eb734dca8f1d72a2e68ebc14b34b3743a317c49e)eaec515
-
Publish JSpark3 v1 card and verified metadata
Verified signaturee9cbbaf -
initial commit
Verified signature279a47b