Instructions to use hkust-aaron/EdgeRock-4B-AWQ with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hkust-aaron/EdgeRock-4B-AWQ with Transformers:

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

pipe = pipeline("text-generation", model="hkust-aaron/EdgeRock-4B-AWQ", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("hkust-aaron/EdgeRock-4B-AWQ", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use hkust-aaron/EdgeRock-4B-AWQ with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "hkust-aaron/EdgeRock-4B-AWQ"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hkust-aaron/EdgeRock-4B-AWQ",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/hkust-aaron/EdgeRock-4B-AWQ

SGLang

How to use hkust-aaron/EdgeRock-4B-AWQ 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 "hkust-aaron/EdgeRock-4B-AWQ" \
    --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": "hkust-aaron/EdgeRock-4B-AWQ",
		"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 "hkust-aaron/EdgeRock-4B-AWQ" \
        --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": "hkust-aaron/EdgeRock-4B-AWQ",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use hkust-aaron/EdgeRock-4B-AWQ with Docker Model Runner:
```
docker model run hf.co/hkust-aaron/EdgeRock-4B-AWQ
```

EdgeRock-4B-AWQ / edgerock_quantization.json

hkust-aaron

Publish calibrated AWQ 4-bit EdgeRock with packed GPU runtime

64f6172 verified 3 days ago

Raw History Blame Contribute Delete

38.3 kB

{
  "method": "awq",
  "bits": 4,
  "group_size": 128,
  "activation_dtype": "bfloat16",
  "scale_dtype": "float32",
  "zero_point": 8,
  "awq_input_scaling": true,
  "modules": {
    "model.layers.0.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.0.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.0.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.0.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.0.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.0.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.1.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.1.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.1.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.1.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.1.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.1.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.2.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.2.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.2.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.2.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.2.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.2.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.self_attn.q_proj": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.self_attn.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.self_attn.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.self_attn.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.memory.q_proj": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.memory.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.memory.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.memory.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.memory.write_gate_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.memory.write_up_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.3.memory.write_down_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.4.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.4.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.4.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.4.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.4.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.4.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.5.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.5.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.5.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.5.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.5.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.5.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.6.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.6.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.6.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.6.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.6.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.6.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.self_attn.q_proj": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.self_attn.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.self_attn.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.self_attn.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.memory.q_proj": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.memory.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.memory.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.memory.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.memory.write_gate_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.memory.write_up_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.7.memory.write_down_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.8.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.8.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.8.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.8.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.8.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.8.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.9.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.9.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.9.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.9.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.9.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.9.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.10.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.10.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.10.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.10.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.10.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.10.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.self_attn.q_proj": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.self_attn.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.self_attn.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.self_attn.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.memory.q_proj": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.memory.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.memory.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.memory.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.memory.write_gate_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.memory.write_up_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.11.memory.write_down_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.12.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.12.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.12.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.12.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.12.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.12.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.13.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.13.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.13.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.13.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.13.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.13.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.14.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.14.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.14.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.14.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.14.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.14.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.self_attn.q_proj": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.self_attn.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.self_attn.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.self_attn.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.memory.q_proj": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.memory.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.memory.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.memory.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.memory.write_gate_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.memory.write_up_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.15.memory.write_down_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.16.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.16.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.16.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.16.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.16.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.16.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.17.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.17.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.17.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.17.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.17.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.17.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.18.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.18.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.18.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.18.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.18.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.18.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.self_attn.q_proj": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.self_attn.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.self_attn.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.self_attn.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.memory.q_proj": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.memory.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.memory.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.memory.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.memory.write_gate_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.memory.write_up_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.19.memory.write_down_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.20.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.20.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.20.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.20.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.20.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.20.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.21.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.21.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.21.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.21.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.21.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.21.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.22.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.22.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.22.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.22.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.22.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.22.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.self_attn.q_proj": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.self_attn.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.self_attn.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.self_attn.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.memory.q_proj": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.memory.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.memory.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.memory.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.memory.write_gate_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.memory.write_up_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.23.memory.write_down_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.24.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.24.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.24.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.24.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.24.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.24.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.25.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.25.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.25.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.25.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.25.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.25.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.26.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.26.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.26.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.26.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.26.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.26.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.self_attn.q_proj": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.self_attn.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.self_attn.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.self_attn.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.memory.q_proj": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.memory.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.memory.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.memory.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.memory.write_gate_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.memory.write_up_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.27.memory.write_down_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.28.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.28.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.28.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.28.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.28.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.28.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.29.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.29.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.29.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.29.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.29.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.29.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.30.linear_attn.out_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.30.linear_attn.in_proj_qkv": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.30.linear_attn.in_proj_z": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.30.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.30.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.30.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.self_attn.q_proj": {
      "shape": [
        8192,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.self_attn.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.self_attn.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.self_attn.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.mlp.gate_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.mlp.up_proj": {
      "shape": [
        9216,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.mlp.down_proj": {
      "shape": [
        2560,
        9216
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.memory.q_proj": {
      "shape": [
        4096,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.memory.k_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.memory.v_proj": {
      "shape": [
        1024,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.memory.o_proj": {
      "shape": [
        2560,
        4096
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.memory.write_gate_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.memory.write_up_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    },
    "model.layers.31.memory.write_down_proj": {
      "shape": [
        2560,
        2560
      ],
      "input_scale": true,
      "bias": false
    }
  }
}