LOOM TAPESTRY 2 — UPLOAD STEPS ============================== 1. Add your two images to this folder FIRST, named exactly: banner.jpg logo.jpg (prompts are in TAPESTRY-2-ARTWORK-PROMPTS.md in the project folder. The name must be .jpg — a .jpeg mismatch broke the images on an earlier release.) 2. On huggingface.co, create a NEW model repo: textilelabs/Loom-Tapestry-2 Owner: textilelabs · Public · License: MIT 3. Files and versions -> Add file -> Upload files. Drag in ALL the CONTENTS of this folder (not the folder itself). Do NOT upload UPLOAD-STEPS.txt — it is for you, not the repo. 4. Commit. That is it — no settings to change. The tags, widget examples and licence all come from the top of README.md. 5. Verify (takes a minute): ollama run hf.co/textilelabs/Loom-Tapestry-2 "who are you" Expected: "Loom Tapestry 2, a small model by Textile Labs." or similar. If Ollama says it cannot find a template, the template/params files did not upload — re-add just those two. 6. Optional: add it to your "Loom-Spark" collection so it sits with the family. WHAT IS IN HERE (14 files + your 2 images = 16) ----------------------------------------------- README.md the model card (tags/widgets are in its header) config.json model architecture model.safetensors the weights, 87 MB generation_config.json tokenizer.json custom 4,096-token BPE tokenizer_config.json special_tokens_map.json loom-tapestry-2-f16.gguf 44 MB, for Ollama / llama.cpp template Ollama reads this automatically (defaults tools:off) params Ollama reads this automatically Modelfile only needed for `ollama create` locally harness.py runnable harness — does the lookups ATTRIBUTION.md licence credits, required by the corpora LICENSE MIT banner.jpg <- you add logo.jpg <- you add