qwen3-0-6b-awq@0.1.0

Immutable release / 0.1.0

PlannedPlanning blockedNot measured

Qwen3-0.6B · AWQ W4A16

A pinned AWQ smoke recipe with explicit calibration and evaluation data, assistant-continuation loss v2, and a declared 16 GiB CUDA target. Execution acceptance remains blocked.

README.md

Before you use this recipe

The smoke recipe pins model, upstream quantizer and both datasets. Adapter-source review and runtime acceptance remain outstanding. The published validator accepts the authored schema. Preparation remains blocked on the adapter source binding and real runtime acceptance; no accepted executable plan is claimed.

No quantized artifact has been produced or published. No accuracy, latency, throughput, memory, or hardware evidence bundle exists.

Model

Model	`Qwen/Qwen3-0.6B`
Family	Qwen3
Revision	`c1899de289a04d12100db370d81485cdf75e47ca`

Compression method

This recipe specifies AWQ with 4-bit weights and 16-bit activations (W4A16), using a group size of 128. The export format is compressed-tensors, targeting vLLM inference.

Ordered stages

quantize
Recipe plugin
lazybrick.llm-compressor-awq@0.13.0
Upstream implementation
https://github.com/vllm-project/llm-compressorde46bfd53513aa87571a8b056a06aeaa5da1c69c
Stage parameters
```
{
  "weight_bits": 4,
  "group_size": 128,
  "symmetric": false,
  "targets": [
    "Linear"
  ],
  "ignore": [
    "lm_head"
  ]
}
```

These pins identify authored declarations. The Hub has not verified the adapter package, installed source, or execution binding.

Numerical semantics

quantize: Unspecified

This recipe pins an implementation but does not declare a numerical profile.

Conformance: Not verified

A declaration is not evidence that an implementation, artifact, or runtime conforms.

Requirements

Accelerator	NVIDIA CUDA
Declared memory target	16 GiB
Runtime	vLLM
Validator	lazybrick 0.0.1
Identity command	`lazybrick fingerprint`

Compatibility is declared, not measured.

Calibration and evaluation

Calibration data	`hf-dataset://HuggingFaceH4/ultrachat_200k` `8049631c405ae6576f93f445c6b8166f76f5505a`
Calibration samples	32
Evaluation data	`hf-dataset://HuggingFaceH4/ultrachat_200k` `8049631c405ae6576f93f445c6b8166f76f5505a`
Evaluation protocol	`lazybrick.ultrachat-assistant-continuation-loss.v2`
Evaluation sample limit	8

Protocol and sample limits describe the intended workload; no measured result is implied.

Use this recipe

Download qwen3-0-6b-awq-0.1.0.recipe.yaml using the download button above. The Usage tab has copyable CLI and Python examples for this exact file. Review the Blockers tab before running it, and inspect or copy the full source in Recipe YAML.

Recipe identity

Recipe	qwen3-0-6b-awq
Version	0.1.0
Recipe digest	`56d0a9cb7d51b3219b906fb6ad8812f387c97b9bb56c3d8b87862a5aea694f66`
File SHA-256	`28d56fd7628f0af97294fc10dd7cfee265a4230d2ab93eb7f89195e22861cc2e`

The recipe digest identifies canonical authored content; the file hash identifies the exact downloaded bytes. Neither is a plan, artifact, run, or evidence identity. Presentation metadata is available separately in entry.json.

Validation

Blocking decisions

stages[0].pluginunsupported_capability

The source smoke adapter is not provided by the published validator CLI. Review the core transport manifest and reconcile its first-party source pin with the reviewed adapter build before execution. Metadata/version agreement is not installed-source attestation.

export.runtimeunsupported_capability

Tokenizer boundary handling, vLLM loading, CUDA execution and allocating-process resource sampling remain unverified acceptance gates. Review the pinned smoke workflow and obtain approval for a GPU run before claiming runtime support.

Usage

Use the exact recipe

These examples use the downloaded file qwen3-0-6b-awq-0.1.0.recipe.yaml.

Install the recorded validator, then run these commands in the download directory. Installation may access the package registry; validation and fingerprinting run locally without model downloads.

Install CLI · uv

uv tool install 'lazybrick==0.0.1'

For the Python example, use uv add 'lazybrick==0.0.1' in your project. With pip, use python -m pip install 'lazybrick==0.0.1'.

CLI · schema validation

lazybrick validate qwen3-0-6b-awq-0.1.0.recipe.yaml

Expected: exit code 0 and a valid-recipe message containing the digest below. This checks the published validator’s schema boundary; it does not resolve references, validate execution compatibility, or clear planning blockers.

Authored recipe identity

lazybrick fingerprint qwen3-0-6b-awq-0.1.0.recipe.yaml

Expected SHA-256: 56d0a9cb7d51b3219b906fb6ad8812f387c97b9bb56c3d8b87862a5aea694f66

The recorded lazybrick 0.0.1 release provides validation and fingerprinting. Planning requires a separately reviewed core checkout, target and plugin manifests, immutable inputs, and verified execution bindings. Review the planning blockers before preparing execution.

Python

from lazybrick import load_recipe

recipe = load_recipe("qwen3-0-6b-awq-0.1.0.recipe.yaml")
print(recipe.digest)

Source

Authored recipe

qwen3-0-6b-awq-0.1.0.recipe.yaml

schema_version: "0.1"

model:
  uri: hf://Qwen/Qwen3-0.6B
  revision: c1899de289a04d12100db370d81485cdf75e47ca
  trust_remote_code: false

stages:
  - id: quantize
    plugin: lazybrick.llm-compressor-awq
    plugin_version: "0.13.0"
    implementation:
      git: https://github.com/vllm-project/llm-compressor
      commit: de46bfd53513aa87571a8b056a06aeaa5da1c69c
    parameters:
      weight_bits: 4
      group_size: 128
      symmetric: false
      targets:
        - Linear
      ignore:
        - lm_head

calibration:
  dataset:
    uri: hf-dataset://HuggingFaceH4/ultrachat_200k
    revision: 8049631c405ae6576f93f445c6b8166f76f5505a
  samples: 32
  seed: 42
  preprocessing_id: lazybrick.qwen-chat-template.v1
  max_sequence_length: 2048

export:
  format: compressed-tensors
  runtime: vllm

target:
  accelerator_family: nvidia-cuda
  min_compute_capability: "8.0"
  device_count: 1
  min_memory_gib: 16
  runtime: vllm

evaluation:
  protocol_id: lazybrick.ultrachat-assistant-continuation-loss.v2
  dataset:
    uri: hf-dataset://HuggingFaceH4/ultrachat_200k
    revision: 8049631c405ae6576f93f445c6b8166f76f5505a
  max_samples: 8
  seed: 1234