Instructions to use normalcomputing/extended-mind-llama-2-7b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use normalcomputing/extended-mind-llama-2-7b with Transformers:

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

pipe = pipeline("text-generation", model="normalcomputing/extended-mind-llama-2-7b", trust_remote_code=True)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("normalcomputing/extended-mind-llama-2-7b", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use normalcomputing/extended-mind-llama-2-7b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "normalcomputing/extended-mind-llama-2-7b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "normalcomputing/extended-mind-llama-2-7b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/normalcomputing/extended-mind-llama-2-7b

SGLang

How to use normalcomputing/extended-mind-llama-2-7b 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 "normalcomputing/extended-mind-llama-2-7b" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "normalcomputing/extended-mind-llama-2-7b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "normalcomputing/extended-mind-llama-2-7b" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "normalcomputing/extended-mind-llama-2-7b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use normalcomputing/extended-mind-llama-2-7b with Docker Model Runner:
```
docker model run hf.co/normalcomputing/extended-mind-llama-2-7b
```

extended-mind-llama-2-7b

File size: 1,445 Bytes

{
  "_name_or_path": "normalcomputing/extended-mind-llama-2-7b",
  "architectures": [
    "ExtendedLlamaForCausalLM"
  ],
  "auto_map": {     
	"AutoConfig": "configuration.ExtendedLlamaConfig",     
	"AutoModel": "modeling.ExtendedLlamaModel",
	"AutoModelForCausalLM": "modeling.ExtendedLlamaForCausalLM"   
  },
  "bos_token_id": 1,
  "eos_token_id": 2,
  "hidden_act": "silu",
  "hidden_size": 4096,
  "initializer_range": 0.02,
  "intermediate_size": 11008,
  "mask_by_sim": false,
  "max_position_embeddings": 4096,
  "memory_device": "cpu",
  "memory_type": "manual",
  "model_type": "extended-llama",
  "num_attention_heads": 32,
  "num_hidden_layers": 32,
  "num_key_value_heads": 32,
  "pretraining_tp": 1,
  "remove_special_ids": true,
  "rms_norm_eps": 1e-05,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "sim_threshold": 0.25,
  "tie_word_embeddings": false,
  "tokenizer_all_special_ids": [
    0,
    1,
    2
  ],
  "topk": 10,
  "torch_dtype": "float32",
  "transformers_version": "4.33.0",
  "use_cache": true,
  "use_external_mind": true,
  "use_external_mind_by_layer": [
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true
  ],
  "vocab_size": 32000
}