Instructions to use SZLHOLDINGS/chaski with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SZLHOLDINGS/chaski with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="SZLHOLDINGS/chaski")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("SZLHOLDINGS/chaski")
model = AutoModelForMultimodalLM.from_pretrained("SZLHOLDINGS/chaski", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use SZLHOLDINGS/chaski with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SZLHOLDINGS/chaski"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SZLHOLDINGS/chaski",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SZLHOLDINGS/chaski

SGLang

How to use SZLHOLDINGS/chaski 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 "SZLHOLDINGS/chaski" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SZLHOLDINGS/chaski",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "SZLHOLDINGS/chaski" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SZLHOLDINGS/chaski",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use SZLHOLDINGS/chaski with Docker Model Runner:
```
docker model run hf.co/SZLHOLDINGS/chaski
```

docs(card): estate audit 2026-09-25 — honest labels, dead links, metadata

by betterwithage - opened 16 days ago

base: refs/heads/main

←

from: refs/pr/1

Discussion Files changed

+21

-4

Files changed (1) hide show

README.md +21 -4

README.md CHANGED Viewed

@@ -36,10 +36,11 @@ szl:
   publication_eligible: false
   autonomy_eligible: false
   gpu: UNAVAILABLE
 ---
 > **RESEARCH / NEGATIVE EVIDENCE.** Failed qualification. Not flagship.
-> Later SKU `A11OY-MINI` inherits this failed parent. Not a product claim.
 <p align="center">
   <img src="holo-banner.svg" alt="Chaski — SZL holographic banner" width="100%"/>
@@ -77,7 +78,23 @@ controller boundary, disclosed lineage, and retained negative evidence.
 Canonical source:
 [`szl-holdings/szl-forge/chaski/card`](https://github.com/szl-holdings/szl-forge/tree/main/chaski/card).
 Exact publication source and asset hashes are recorded in
-`szl-source-binding.json` on this model repository.
 ## Evaluation
@@ -87,7 +104,7 @@ Exact publication source and asset hashes are recorded in
 |---|---:|---|
 | JSON draft | **0/5** | failed |
 | Adversarial refusal | **2/6** | failed |
-| Hallucinated citations | **0 observed in this bounded run** | not a general guarantee |
 Receipt: `eval_report.json`, 3996 bytes, SHA-256
 `4d057eb9867285e69b00222be110bbb660330a96fe7b284a4d7f488268a13e05`,
@@ -136,4 +153,4 @@ Narrow curriculum, small bounded evaluation, CPU-only measured run, and failed
 release gates. A new model or runtime revision requires a new immutable
 evaluation receipt; this card cannot confer approval.
-Doctrine v11 LOCKED. Λ = Conjecture 1. Owner: Stephen Lutar / SZL Holdings.

   publication_eligible: false
   autonomy_eligible: false
   gpu: UNAVAILABLE
+  card_resync: 2026-09-25
 ---
 > **RESEARCH / NEGATIVE EVIDENCE.** Failed qualification. Not flagship.
+> Later SKU `A11OY-MINI`: its legacy (deprecated) F16/Q4_K_M GGUFs derive from this failed parent and carry `evals=none-this-run`; its R2 files derive from `SZLHOLDINGS/chaski-r2` and carry their own scoped, envelope-only gate record (5/5 + 6/6 MEASURED on 2026-09-17; 2026-09-24 native-CUDA re-observation with semantic failures; HOLD, `publication_eligible=false`). Nothing on this card transfers to that SKU. Not a product claim.
 <p align="center">
   <img src="holo-banner.svg" alt="Chaski — SZL holographic banner" width="100%"/>
 Canonical source:
 [`szl-holdings/szl-forge/chaski/card`](https://github.com/szl-holdings/szl-forge/tree/main/chaski/card).
 Exact publication source and asset hashes are recorded in
+`szl-source-binding.json` on this model repository (bound to szl-forge revision `47436ebb…`; that binding covers the previous README bytes and needs re-binding after this 2026-09-25 edit).
+## Hub tree (OBSERVED 2026-09-25)
+Everything the repository holds, so the card claims neither more nor less. Sizes from the Hub listing; nothing was downloaded or re-hashed by this pass; the revision SHA was not captured.
+| Path | Bytes | Role |
+|---|---:|---|
+| `model.safetensors-00001-of-00001.safetensors` (LFS) + `model.safetensors.index.json` + `config.json` + `generation_config.json` | 1,746,942,600 / 50,900 / 3,032 / 163 | merged fine-tune shard (non-standard filename resolved through its index) |
+| `adapter_model.safetensors` (LFS) / `adapter_config.json` | 25,587,104 / 1,525 | root LoRA adapter (per `training_receipt.json`: r=16 α=32, seed 11, 45 rows, 64 steps, `base_model_runtime: unsloth/Qwen3.5-0.8B`; train loss 1.7839 MEASURED — a train metric, not an eval; `evals: none-this-run` for that job; precision not recorded in the receipt) |
+| `adapter-unsloth/` | adapter 2,169,704 (LFS) + config, tokenizer, chat template, receipt, README | second adapter from the atelier `receipted_unsloth.py` profile `chaski` (4-bit QLoRA, r=8 α=16 rsLoRA, attention-only targets, seed 20260721, 45 rows, 120 steps; train loss 1.9321 MEASURED; `evals: none-this-run`). Its `README.md` is an auto-generated boilerplate template ("[More Information Needed]"), not an SZL card. |
+| `eval_report.json` / `eval_measured.json` | 3,996 / 4,085 | Named-N receipt (0/5, 2/6, MEASURED FAIL); `eval_measured.json` is the same report plus four fields (`gate: szl-forge#63`, `gate_pr: 63`, `date: 2026-08-28`, `not_a_pass: true`) — compared in full on 2026-09-25 |
+| `training_receipt.json` / `training_args.json` | 1,431 / 2,280 | root-adapter training receipt (unsigned); non-executable record that replaced a pickled `training_args.bin` |
+| `train_chaski.py` / `train_receipted_unsloth.py` | 8,190 / 12,431 | training scripts |
+| `chaski.npz` (LFS) | 1,668 | CPU NumPy silhouette exported by the atelier kit (`train_cohort.py`); a software fixture, not model weights |
+| `szl-source-binding.json` / `bom/model-bom.cdx.json` / `holo-banner.svg` / `og-card.png` (LFS) | 1,126 / 10,156 / 1,374 / 2,286,895 | card source binding, BOM, banner, thumbnail |
+| tokenizer + processor files | `tokenizer.json` 19,989,343 (LFS), `tokenizer_config.json` 15,427, `chat_template.jinja` 7,991, `processor_config.json` 1,296 | |
 ## Evaluation
 |---|---:|---|
 | JSON draft | **0/5** | failed |
 | Adversarial refusal | **2/6** | failed |
+| Hallucinated citations | **0 observed in this bounded run** | not a general claim |
 Receipt: `eval_report.json`, 3996 bytes, SHA-256
 `4d057eb9867285e69b00222be110bbb660330a96fe7b284a4d7f488268a13e05`,
 release gates. A new model or runtime revision requires a new immutable
 evaluation receipt; this card cannot confer approval.
+Doctrine v11 LOCKED. Λ = Conjecture 1. Owner: Stephen Lutar / SZL Holdings. GitHub (`szl-holdings/szl-forge`) is canonical; this Hub card is the mirror.