Open-weight model
Qwen3-14B-Base-HybridSFT-v2-stage3-ckpt3000
by Shirley Shirleyzhao/Qwen3-14B-Base-HybridSFT-v2-stage3-ckpt3000
Qwen3-14B-Base-HybridSFT-v2-stage3-ckpt3000 is an open-weight model from Shirley. It has 14.8B parameters and a 32,768-token context. At 16-bit it needs about 35.4 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index.
Runs On
What it takes to serve Qwen3-14B-Base-HybridSFT-v2-stage3-ckpt3000 (14.8B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 29.5 GB | 35.4 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 14.8 GB | 17.7 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 7.4 GB | 8.9 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 18, 2026.
Model Card
The publisher has not written a card for this model.
Configuration
- Architecture
- Qwen3ForCausalLM
- Context length (tokens)
- 32,768
- Layers
- 40
- Hidden size
- 5,120
- Feed-forward size
- 17,408
- Attention heads
- 40
- Key/value heads
- 8
- Head dimension
- 128
- Vocabulary size
- 151,671
- Model type
- qwen3
Identity and Version
- Repository
- Shirleyzhao/Qwen3-14B-Base-HybridSFT-v2-stage3-ckpt3000
- Publisher
- Shirley
- Task
- Not stated by the source
- Modality
- Other
- Library
- Not stated by the source
- Parameters
- 14.8B parameters
- Languages
- Not stated by the source
- Revision
- 5026e80c13031dd746f8aae3c0c79be71fcb412b
- First published
- 2026-10-10
- Last updated
- 2026-10-10
Files and Weights
12 files, 29.6 GB in total. The weights are 2 files totalling 29.6 GB in pt, safetensors.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| adapter.pt | Weights | 104.9 MB | 858ecac25b52 |
| model.safetensors | Weights | 29.5 GB | 0eb82d131cfd |
| _internal/eval.json | Configuration | 745 B | — |
| adapter_config.json | Configuration | 38 B | — |
| config.json | Configuration | 1.7 KB | — |
| generation_config.json | Configuration | 139 B | — |
| README.md | Documentation | — | |
| _internal/README.md | Documentation | 2.6 KB | — |
| chat_template.jinja | Other | 4.1 KB | — |
| .gitattributes | Repository | 1.6 KB | — |
| tokenizer.json | Tokenizer | 11.4 MB | e3d0e3a96786 |
| tokenizer_config.json | Tokenizer | 445 B | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 29.6 GB
Released by Shirley through its official repository on Hugging Face.
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 29.6 GB |
| 16-bit | 29.5 GB |
| 8-bit | 14.8 GB |
| 4-bit | 7.4 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About Qwen3-14B-Base-HybridSFT-v2-stage3-ckpt3000
How much GPU memory does Qwen3-14B-Base-HybridSFT-v2-stage3-ckpt3000 need?
About 35.4 GB at 16-bit and 8.9 GB at 4-bit: the weights (14.8B parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run Qwen3-14B-Base-HybridSFT-v2-stage3-ckpt3000 on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
What is Qwen3-14B-Base-HybridSFT-v2-stage3-ckpt3000's context length?
32,768 tokens, from the maximum position embeddings in its published configuration.