Upload PowerServe/Llama-3.2-1B-PowerServe-QNN29-8G3

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+ggml/weights.gguf filter=lfs diff=lfs merge=lfs -text
+vocab.gguf filter=lfs diff=lfs merge=lfs -text

ggml/weights.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a67e6392c8db79112d70a3122b10775be3fa69e261e37122c73901810cb41295
+size 703209312

model.json

@@ -0,0 +1,26 @@
+{
+    "model_id": "llama_3.2_1b",
+    "model_arch": "llama",
+    "version": 20241202,
+    "llm_config": {
+        "embed_dim": 2048,
+        "ffn_dim": 8192,
+        "head_size": 64,
+        "kv_dim": 512,
+        "n_attn_heads": 32,
+        "n_attn_kv_heads": 8,
+        "n_ctx": 131072,
+        "n_layers": 16,
+        "norm_eps": 9.999999747378752e-06,
+        "vocab_size": 128256,
+        "rope_config": {
+            "n_rope_ctx_orig": 131072,
+            "rope_attn_factor": 1.0,
+            "rope_dim": 64,
+            "rope_freq_base": 500000.0,
+            "rope_freq_scale": 1.0,
+            "rope_scale_type": "linear",
+            "rope_type": 0
+        }
+    }
+}

qnn/config.json

@@ -0,0 +1,69 @@
+{
+    "model_parameters": {
+        "n_layers": 16,
+        "vocab_size": 128256,
+        "embed_dim": 2048,
+        "ffn_hidden_dim": 8192,
+        "head_dim": 64,
+        "n_kv_heads": 8,
+        "rope_theta": 500000.0,
+        "rms_norm_eps": 1e-05,
+        "attention_mask_value": -100000.0,
+        "tie_embedding": true
+    },
+    "qnn_parameters": {
+        "n_hvx_threads": 4
+    },
+    "graphs": [
+        {
+            "type": "transformers",
+            "start_layer_id": 0,
+            "end_layer_id": 16,
+            "batch_size": 1,
+            "context_size": 2048,
+            "cache_size": 1920,
+            "graph_name": "batch_1",
+            "model_path": "llama3_2_1b.bin",
+            "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
+            "kv_size": 13,
+            "x_name": "x",
+            "out_name": "out"
+        },
+        {
+            "type": "transformers",
+            "start_layer_id": 0,
+            "end_layer_id": 16,
+            "batch_size": 128,
+            "context_size": 2048,
+            "cache_size": 1920,
+            "graph_name": "batch_128",
+            "model_path": "llama3_2_1b.bin",
+            "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
+            "kv_size": 13,
+            "x_name": "x",
+            "out_name": "out"
+        }
+    ],
+    "embeddings": [
+        {
+            "graph_name": "batch_1",
+            "model_path": "lm_head.bin",
+            "batch_size": 1,
+            "input_type": 562,
+            "input_dim": 2048,
+            "output_dim": 128256,
+            "x_name": "x",
+            "out_name": "logits"
+        },
+        {
+            "graph_name": "batch_128",
+            "model_path": "lm_head.bin",
+            "batch_size": 128,
+            "input_type": 562,
+            "input_dim": 2048,
+            "output_dim": 128256,
+            "x_name": "x",
+            "out_name": "logits"
+        }
+    ]
+}