Instructions to use ewald1976/Corridor-D-RevC-12B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ewald1976/Corridor-D-RevC-12B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="ewald1976/Corridor-D-RevC-12B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ewald1976/Corridor-D-RevC-12B")
model = AutoModelForCausalLM.from_pretrained("ewald1976/Corridor-D-RevC-12B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use ewald1976/Corridor-D-RevC-12B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ewald1976/Corridor-D-RevC-12B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ewald1976/Corridor-D-RevC-12B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ewald1976/Corridor-D-RevC-12B

SGLang

How to use ewald1976/Corridor-D-RevC-12B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "ewald1976/Corridor-D-RevC-12B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ewald1976/Corridor-D-RevC-12B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "ewald1976/Corridor-D-RevC-12B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ewald1976/Corridor-D-RevC-12B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ewald1976/Corridor-D-RevC-12B with Docker Model Runner:
```
docker model run hf.co/ewald1976/Corridor-D-RevC-12B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Corridor-D-RevC-12B

Modelcard: ewald1976/Corridor-D-RevC-12B

Description: Corridor-D-RevC-12B s a revision of Corridor-D including Corridor-C, incorporating feedback from deployment. The model maintains the core behavioral traits of its predecessor while addressing known issues in memory usage and response latency.

Architecture: Corridor-D builds on the modular architecture of Corridor-C, with additional layers for sensory processing and motor control. Key improvements include:

Memory optimizations: Reorganized data structures to reduce cache misses.
Latency reduction: Refactored communication pathways between modules.
Noise resilience: Added error-checking routines at critical junctures.

Behavior: The model retains the core behavioral traits of Corridor-C, including:

Navigation through unfamiliar environments
Object recognition and manipulation
Adaptive decision-making based on environmental cues

New behaviors introduced in RevC include:

Improved obstacle avoidance using depth perception.
Enhanced prioritization of critical tasks under stress conditions.

Deployment Status: Corridor-D is currently in limited deployment, with performance metrics indicating improvement over previous versions. Feedback from early users suggests reduced cognitive load and increased reliability.

Limitations: Known limitations include potential issues with high-precision spatial reasoning and occasional hesitation in ambiguous situations.

Roadmap: Future revisions will focus on integrating recent advances in attention modeling and improving energy efficiency during idle states.

This is a merge of pre-trained language models created using mergekit.

Configuration

The following YAML configuration was used to produce this model:

# corridor-d-revc-12b.yaml
models:
  - model: ewald1976/Corridor-D-12B
  - model: ewald1976/Corridor-C-12B

merge_method: slerp
base_model: ewald1976/Corridor-D-12B

parameters:
  t:
    - filter: self_attn
      value: 0.25
    - filter: mlp
      value: 0.35
    - value: 0.30

dtype: bfloat16

Settings

temp: 0.6-0.7
rep pen: 1.05
min_p: 0.05
top_p: 0.90-0.95

This model is provided as-is, without warranty of any kind. The creator of this merge accepts no responsibility for any content generated by the model, including harmful, offensive, misleading, or otherwise objectionable outputs. You use it at your own risk.