Downloads · 30 days
25
8% of all-time downloads
mariana-qwen3-14b
mmoralesf/mariana-qwen3-14b
mariana-qwen3-14b is a text generation model from mmoralesf. Use it when you need the model to write or continue text. The card lists the license as apache-2.0.
Fine-tuned version of Qwen3-14B for automotive sales conversations in Mexican Spanish.
All-time downloads
304
Public
Parameters
14.8B
40.1 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors29.5 GB · 74%
At a glance
- Task
- Text Generation
- License
- apache-2.0
- Model type
- qwen3
- Access
- Public
- Created
- Feb 14, 2026
- Updated
- Feb 14, 2026
- SHA
- 609d57ea
Base models
Model card
From the Hugging Face model README
Mariana — Qwen3-14B Fine-tuned (Automotive Sales Assistant)
Fine-tuned version of Qwen3-14B for automotive sales conversations in Mexican Spanish.
Model Details
- Base model: Qwen/Qwen3-14B
- Fine-tuning: LoRA + merged
- Language: Spanish (Mexico)
- Domain: Automotive sales (used cars)
- Tool calling: Hermes format, 11 custom tools
Intended Use
Virtual sales assistant for Autos TREFA, a used car dealership in Mexico. Handles:
- Vehicle inventory search and recommendations
- Financing calculations
- Business information (hours, locations, policies)
- Lead capture and follow-up
Available Formats
| Format | Path | Use Case |
|---|---|---|
| Safetensors (merged) | / | vLLM, transformers, TGI |
| GGUF Q5_K_M | /gguf/ | llama.cpp, Ollama |
Usage with vLLM
vllm serve marianomorales/mariana-qwen3-14b \
--max-model-len 8192 \
--enable-auto-tool-choice \
--tool-call-parser hermes
Training
- Method: LoRA (r=16, alpha=32)
- Dataset: Synthetic automotive sales conversations
- Hardware: NVIDIA RTX 5880 Ada (48GB VRAM)
- Task
- Text Generation
- Type
- qwen3
- License
- apache-2.0
- Languages
- es
- Created
- Feb 14, 2026
- Updated
- Feb 14, 2026