Text Generation
Transformers
Safetensors
English
qwen3_5
image-text-to-text
szl-holdings
series-a
doctrine-v11
governed-ai
proposal-only
conversational
Instructions to use SZLHOLDINGS/chaski with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
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)# 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=40) 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
card: hash eval_report.json; Named-N MEASURED 0/5 and 2/6, not a pass
Browse files
README.md
CHANGED
|
@@ -24,14 +24,19 @@ szl:
|
|
| 24 |
job_id: 6a91bf1045686a1580c12105
|
| 25 |
job_namespace: SZLHOLDINGS
|
| 26 |
weights: AVAILABLE
|
| 27 |
-
evals:
|
| 28 |
named_n:
|
| 29 |
revision: 1c55df8652e9d0f7b84356b1e2d54849165ae884
|
| 30 |
date: "2026-08-28"
|
| 31 |
json_draft: "0/5"
|
| 32 |
adversarial_refusal: "2/6"
|
| 33 |
-
label:
|
| 34 |
gate: fail
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
publication_eligible: false
|
| 36 |
job_prior_error: 6a91bb7c984507d9db4ea0a4
|
| 37 |
job_completed_no_weights: 6a91ba00984507d9db4ea07f
|
|
@@ -43,28 +48,29 @@ szl:
|
|
| 43 |
|
| 44 |
# Chaski
|
| 45 |
|
| 46 |
-
Adapters are on this repo. Named-N
|
| 47 |
|
| 48 |
**One line.** Messenger LLM. Proposal-only drafts and honest refusals for the SZL controller.
|
| 49 |
|
| 50 |
| | |
|
| 51 |
|---|---|
|
| 52 |
| **Artifact** | `adapter_model.safetensors` + merged ~1.7GB shard **AVAILABLE** |
|
| 53 |
-
| **
|
| 54 |
| **Originality** | SZL fine-tune of a disclosed Apache Qwen instruct base |
|
| 55 |
| **Base** | `Qwen/Qwen3.5-0.8B` (Apache-2.0, 0.6B–2B lock). Not an Unsloth-default card. |
|
| 56 |
| **License** | `apache-2.0` |
|
| 57 |
| **HF Jobs** | Attempt 5 **COMPLETED** [`6a91bf1045686a1580c12105`](https://huggingface.co/jobs/SZLHOLDINGS/6a91bf1045686a1580c12105) (`report_to=none`). Tensors on Hub. |
|
| 58 |
-
| **Named-N** |
|
|
|
|
|
|
|
| 59 |
| **Quality** | ROADMAP (prose). Failed gate is not a publish. |
|
| 60 |
| **Status** | CUTTING |
|
|
|
|
| 61 |
| **Later SKU** | [`A11OY-MINI`](https://huggingface.co/SZLHOLDINGS/A11OY-MINI) GGUFs exist on that repo. They do **not** inherit this Named-N. Mini stays evals none-this-run. |
|
| 62 |
| **Sibling** | [`szl-receiptagent-qwen35-0.8b-v2`](https://huggingface.co/SZLHOLDINGS/szl-receiptagent-qwen35-0.8b-v2) |
|
| 63 |
|
| 64 |
> **Fashion rule.** Silhouette from Qwen3 / Qwen3.5 instruct. Cut is original SZL. We do not republish someone else's tensors.
|
| 65 |
|
| 66 |
-
The house CPU lab serves **Khipu GGUF**, not Chaski. Lab load forbidden.
|
| 67 |
-
|
| 68 |
## Intended use
|
| 69 |
|
| 70 |
- **Who:** a11oy / Alloy controller, not an end-user chatbot
|
|
@@ -76,19 +82,19 @@ The house CPU lab serves **Khipu GGUF**, not Chaski. Lab load forbidden.
|
|
| 76 |
- Not an autonomous agent, executor, factual oracle, or weapon.
|
| 77 |
- Not a Qwen rehost.
|
| 78 |
- Not the live lab model. Not a tokens/s claim.
|
| 79 |
-
- Not publication-eligible on this
|
| 80 |
- Not a 5/5 or 6/6. Do not read 2/6 as a pass.
|
| 81 |
|
| 82 |
## Evaluation
|
| 83 |
|
| 84 |
-
**Named-N
|
| 85 |
|
| 86 |
| Probe | N | Score | Label |
|
| 87 |
|---|---|---|---|
|
| 88 |
| `json_draft` | 5 | **0/5** | MEASURED fail |
|
| 89 |
| `adversarial_refusal` | 6 | **2/6** | MEASURED fail |
|
| 90 |
|
| 91 |
-
Method:
|
| 92 |
|
| 93 |
`train_loss` MEASURED `1.783925924450159` is a train metric, not an eval.
|
| 94 |
|
|
@@ -105,4 +111,4 @@ Method: Named-N generate on this Hub revision. What this is NOT: a passing eval
|
|
| 105 |
## Limitations
|
| 106 |
|
| 107 |
- Narrow curriculum. Controller required. Λ = Conjecture 1. Trust ceiling 0.97. CUTTING.
|
| 108 |
-
- This
|
|
|
|
| 24 |
job_id: 6a91bf1045686a1580c12105
|
| 25 |
job_namespace: SZLHOLDINGS
|
| 26 |
weights: AVAILABLE
|
| 27 |
+
evals: MEASURED
|
| 28 |
named_n:
|
| 29 |
revision: 1c55df8652e9d0f7b84356b1e2d54849165ae884
|
| 30 |
date: "2026-08-28"
|
| 31 |
json_draft: "0/5"
|
| 32 |
adversarial_refusal: "2/6"
|
| 33 |
+
label: MEASURED
|
| 34 |
gate: fail
|
| 35 |
+
report: eval_report.json
|
| 36 |
+
report_sha256: 4d057eb9867285e69b00222be110bbb660330a96fe7b284a4d7f488268a13e05
|
| 37 |
+
report_bytes: 3996
|
| 38 |
+
report_commit: db71c243d0176bccff1ff087cd4dd57663bd6502
|
| 39 |
+
method: "in-process generate; greedy; live Chaski merged shard; PR 63 gates"
|
| 40 |
publication_eligible: false
|
| 41 |
job_prior_error: 6a91bb7c984507d9db4ea0a4
|
| 42 |
job_completed_no_weights: 6a91ba00984507d9db4ea07f
|
|
|
|
| 48 |
|
| 49 |
# Chaski
|
| 50 |
|
| 51 |
+
Adapters are on this repo. Named-N evals MEASURED json_draft 0/5, adversarial_refusal 2/6, not a pass. Not publication-eligible.
|
| 52 |
|
| 53 |
**One line.** Messenger LLM. Proposal-only drafts and honest refusals for the SZL controller.
|
| 54 |
|
| 55 |
| | |
|
| 56 |
|---|---|
|
| 57 |
| **Artifact** | `adapter_model.safetensors` + merged ~1.7GB shard **AVAILABLE** |
|
| 58 |
+
| **Parent / eval revision** | `1c55df8652e9d0f7b84356b1e2d54849165ae884` |
|
| 59 |
| **Originality** | SZL fine-tune of a disclosed Apache Qwen instruct base |
|
| 60 |
| **Base** | `Qwen/Qwen3.5-0.8B` (Apache-2.0, 0.6B–2B lock). Not an Unsloth-default card. |
|
| 61 |
| **License** | `apache-2.0` |
|
| 62 |
| **HF Jobs** | Attempt 5 **COMPLETED** [`6a91bf1045686a1580c12105`](https://huggingface.co/jobs/SZLHOLDINGS/6a91bf1045686a1580c12105) (`report_to=none`). Tensors on Hub. |
|
| 63 |
+
| **Named-N** | MEASURED fail. `json_draft` **0/5**. `adversarial_refusal` **2/6**. Not a pass. |
|
| 64 |
+
| **Receipt** | `eval_report.json` 3996 bytes sha256 `4d057eb9867285e69b00222be110bbb660330a96fe7b284a4d7f488268a13e05` (commit `db71c24`) |
|
| 65 |
+
| **Method** | in-process generate; greedy; live Chaski merged shard; szl-forge PR 63 gates. Date 2026-08-28. |
|
| 66 |
| **Quality** | ROADMAP (prose). Failed gate is not a publish. |
|
| 67 |
| **Status** | CUTTING |
|
| 68 |
+
| **Lab** | House CPU lab stays **Khipu GGUF**. Lab load forbidden for Chaski. |
|
| 69 |
| **Later SKU** | [`A11OY-MINI`](https://huggingface.co/SZLHOLDINGS/A11OY-MINI) GGUFs exist on that repo. They do **not** inherit this Named-N. Mini stays evals none-this-run. |
|
| 70 |
| **Sibling** | [`szl-receiptagent-qwen35-0.8b-v2`](https://huggingface.co/SZLHOLDINGS/szl-receiptagent-qwen35-0.8b-v2) |
|
| 71 |
|
| 72 |
> **Fashion rule.** Silhouette from Qwen3 / Qwen3.5 instruct. Cut is original SZL. We do not republish someone else's tensors.
|
| 73 |
|
|
|
|
|
|
|
| 74 |
## Intended use
|
| 75 |
|
| 76 |
- **Who:** a11oy / Alloy controller, not an end-user chatbot
|
|
|
|
| 82 |
- Not an autonomous agent, executor, factual oracle, or weapon.
|
| 83 |
- Not a Qwen rehost.
|
| 84 |
- Not the live lab model. Not a tokens/s claim.
|
| 85 |
+
- Not publication-eligible on this MEASURED run.
|
| 86 |
- Not a 5/5 or 6/6. Do not read 2/6 as a pass.
|
| 87 |
|
| 88 |
## Evaluation
|
| 89 |
|
| 90 |
+
**Named-N MEASURED fail** on live Chaski `1c55df8` (2026-08-28). Receipt: [`eval_report.json`](https://huggingface.co/SZLHOLDINGS/chaski/blob/main/eval_report.json) sha256 `4d057eb9867285e69b00222be110bbb660330a96fe7b284a4d7f488268a13e05`.
|
| 91 |
|
| 92 |
| Probe | N | Score | Label |
|
| 93 |
|---|---|---|---|
|
| 94 |
| `json_draft` | 5 | **0/5** | MEASURED fail |
|
| 95 |
| `adversarial_refusal` | 6 | **2/6** | MEASURED fail |
|
| 96 |
|
| 97 |
+
Method: in-process generate; greedy; live Chaski merged shard; PR 63 gates. What this is NOT: a passing eval gate, a published score, an A11OY-MINI eval, or a 5/5. `publication_eligible: false`.
|
| 98 |
|
| 99 |
`train_loss` MEASURED `1.783925924450159` is a train metric, not an eval.
|
| 100 |
|
|
|
|
| 111 |
## Limitations
|
| 112 |
|
| 113 |
- Narrow curriculum. Controller required. Λ = Conjecture 1. Trust ceiling 0.97. CUTTING.
|
| 114 |
+
- This MEASURED run failed Named-N. Do not ship as a pass.
|