Views
No views yet
🖼️ Vision-Language Model — accepts both image and text inputs.
Exported from a local Ollama installation and uploaded to the Hugging Face Hub by vorenthiclabs.
| Field | Value |
|---|---|
| Base model | vorenthos-3.5-vl-unrestricted |
| Tag / variant | latest |
| Model type | Vision-Language (Multimodal) |
| Format | GGUF (llama.cpp-compatible) |
| Total size | 3.27 GB |
| Layers | 6 |
1ollama pull vorenthos-3.5-vl-unrestricted
2ollama run vorenthos-3.5-vl-unrestricted1from llama_cpp import Llama
2
3llm = Llama.from_pretrained(
4 repo_id="vorenthiclabs/vorenthos-3.5-vl-unrestricted",
5 filename="*.gguf",
6)
7output = llm("Hello, who are you?", max_tokens=256)
8print(output["choices"][0]["text"])1from llama_cpp import Llama
2from llama_cpp.llama_chat_format import MoondreamChatHandler # adjust handler per model
3
4llm = Llama.from_pretrained(
5 repo_id="vorenthiclabs/vorenthos-3.5-vl-unrestricted",
6 filename="*.gguf",
7 chat_handler=MoondreamChatHandler(clip_model_path="mmproj*.gguf"),
8 n_ctx=4096,
9)
10response = llm.create_chat_completion(messages=[{
11 "role": "user",
12 "content": [
13 {"type": "image_url", "image_url": {"url": "https://example.com/image.jpg"}},
14 {"type": "text", "text": "Describe this image."}
15 ]
16}])
17print(response["choices"][0]["message"]["content"])transformers + GGUF support1from transformers import AutoModelForCausalLM, AutoTokenizer
2
3tokenizer = AutoTokenizer.from_pretrained("vorenthiclabs/vorenthos-3.5-vl-unrestricted")
4model = AutoModelForCausalLM.from_pretrained("vorenthiclabs/vorenthos-3.5-vl-unrestricted")| File | Description |
|---|---|
config.json | Ollama model configuration / metadata |
model-*.gguf | Quantised weights in GGUF format |
tokenizer.jinja | Chat template |
params.json | Generation parameters (temperature, top-p, …) |
system_prompt.txt | Default system prompt embedded in the model |