Upload 2 files

README.md

@@ -1,3 +1,41 @@
----
-license: llama3.2
----
+---
+license: llama3.2
+inference: false
+base_model: meta-llama/Llama-3.2-3B-Instruct
+base_model_relation: quantized 
+tags:
+- green
+- p3
+- llmware-chat
+- onnx-qnn
+- emerald
+---
+
+# llama-3.2-3b-onnx-qnn
+
+**llama-3.2-3b-onnx-qnn** is an ONNX QNN int4 quantized version of Llama 3.2 3B Instruct, providing a very small, very fast inference implementation, optimized for NPU deployment on Windows ARM64 AI PCs with Snapdragon Elite X NPU processors.    
+
+[**llama-3.2-3b-instruct**](https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct) is a new 3B chat foundation model from Meta. 
+
+
+### Model Description
+
+- **Developed by:** meta-llama
+- **Quantized by:** llmware
+- **Model type:** llama-3.2 
+- **Parameters:** 3 billion
+- **Model Parent:** meta-llama/Meta-Llama-3.2-3B-Instruct  
+- **Language(s) (NLP):** English  
+- **License:** Llama 3.2 Community License
+- **Uses:** General chat use cases 
+- **RAG Benchmark Accuracy Score:** NA  
+- **Quantization:** int4  
+  
+
+## Model Card Contact
+
+[llmware on github](https://www.github.com/llmware-ai/llmware)  
+
+[llmware on hf](https://www.huggingface.co/llmware)  
+
+[llmware website](https://www.llmware.ai)

config.json

@@ -0,0 +1,39 @@
+{
+  "_name_or_path": "meta-llama/Llama-3.2-3B-Instruct",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": [
+    128001,
+    128008,
+    128009
+  ],
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 3072,
+  "initializer_range": 0.02,
+  "intermediate_size": 8192,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 24,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 32.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": true,
+  "transformers_version": "4.45.2",
+  "use_cache": true,
+  "vocab_size": 128256
+}