README.md

---
base_model:
- Qwen/Qwen3-8B
library_name: transformers
license: mit
pipeline_tag: text-generation
---

# Model Card for SubconsciousDev/TIM-8b-preview

TIM is a model that reasons on recursive task trees formatted as JSON structures.

## Model Details

### Model Description

- **Developed by:** MIT and Subconscious
- **Model type:** Structural reasoning model
- **License:** MIT License
- **Finetuned from model [optional]:** Qwen/Qwen3-8b

### Model Sources

- **Repository:** [TIMRUN](https://github.com/subconscious-systems/TIMRUN)
- **Paper:** [Beyond Context Limits: Subconscious Threads for Long-Horizon Reasoning](https://arxiv.org/pdf/2507.16784)
- **Demo:** [Subconscious API platform](https://www.subconscious.dev/)

## Sample Usage

You can use this model with the `transformers` library, leveraging `trust_remote_code=True`.

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

# Load the model and tokenizer
model_name = "SubconsciousDev/TIM-8b-preview"
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.bfloat16, # Use torch.float16 for GPUs that don't support bfloat16
    device_map="auto",
    trust_remote_code=True
)

# Example: Simple text generation
prompt_text = "What is the capital of France?"
input_ids = tokenizer(prompt_text, return_tensors="pt").input_ids.to(model.device)

output_ids = model.generate(input_ids, max_new_tokens=50, do_sample=True, temperature=0.7)
response = tokenizer.decode(output_ids[0], skip_special_tokens=True)

print(response)
```