Upload PowerServe/Llama-3.1-8B-PowerServe-QNN29-8G3

.gitattributes

@@ -33,3 +33,10 @@ 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
+vocab.gguf filter=lfs diff=lfs merge=lfs -text
+ggml/weights.gguf filter=lfs diff=lfs merge=lfs -text
+qnn/lm_head.bin filter=lfs diff=lfs merge=lfs -text
+qnn/llama3_1_8b_2.bin filter=lfs diff=lfs merge=lfs -text
+qnn/llama3_1_8b_3.bin filter=lfs diff=lfs merge=lfs -text
+qnn/llama3_1_8b_1.bin filter=lfs diff=lfs merge=lfs -text
+qnn/llama3_1_8b_0.bin filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,3 @@
+---
+license: apache-2.0
+---

ggml/weights.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b2a3f5861f15178ac12104c8b0206c40f7591358e6cf5bc0f423965df718ac8
+size 4525777760

model.json

@@ -0,0 +1,27 @@
+{
+    "model_id": "llama3.1_8b_8gen3",
+    "model_arch": "llama",
+    "version": 20241202,
+    "llm_config": {
+        "embed_dim": 4096,
+        "ffn_dim": 14336,
+        "head_size": 128,
+        "kv_dim": 1024,
+        "n_attn_heads": 32,
+        "n_attn_kv_heads": 8,
+        "n_ctx": 131072,
+        "n_layers": 32,
+        "norm_eps": 9.999999747378752e-06,
+        "vocab_size": 128256,
+        "rope_config": {
+            "n_rope_ctx_orig": 131072,
+            "rope_attn_factor": 1.0,
+            "rope_dim": 128,
+            "rope_freq_base": 500000.0,
+            "rope_freq_scale": 1.0,
+            "rope_scale_type": "linear",
+            "rope_type": 0
+        }
+    },
+    "vision": {}
+}

qnn/config.json

@@ -0,0 +1,147 @@
+{
+  "model_parameters": {
+    "n_layers": 32,
+    "vocab_size": 128256,
+    "embed_dim": 4096,
+    "ffn_hidden_dim": 14336,
+    "head_dim": 128,
+    "n_kv_heads": 8,
+    "rope_theta": 500000.0,
+    "rms_norm_eps": 1e-05,
+    "attention_mask_value": -100000.0,
+    "tie_embedding": false
+  },
+  "qnn_parameters": {
+    "n_hvx_threads": 4
+  },
+  "graphs": [
+    {
+      "type": "transformers",
+      "start_layer_id": 0,
+      "end_layer_id": 8,
+      "batch_size": 1,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_1",
+      "model_path": "llama3_1_8b_0.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": 8,
+      "end_layer_id": 16,
+      "batch_size": 1,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_1",
+      "model_path": "llama3_1_8b_1.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": 16,
+      "end_layer_id": 24,
+      "batch_size": 1,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_1",
+      "model_path": "llama3_1_8b_2.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": 24,
+      "end_layer_id": 32,
+      "batch_size": 1,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_1",
+      "model_path": "llama3_1_8b_3.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": 8,
+      "batch_size": 128,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_128",
+      "model_path": "llama3_1_8b_0.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": 8,
+      "end_layer_id": 16,
+      "batch_size": 128,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_128",
+      "model_path": "llama3_1_8b_1.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": 16,
+      "end_layer_id": 24,
+      "batch_size": 128,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_128",
+      "model_path": "llama3_1_8b_2.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": 24,
+      "end_layer_id": 32,
+      "batch_size": 128,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_128",
+      "model_path": "llama3_1_8b_3.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,
+      "x_name": "x",
+      "out_name": "logits"
+    },
+    {
+      "graph_name": "batch_128",
+      "model_path": "lm_head.bin",
+      "batch_size": 128,
+      "x_name": "x",
+      "out_name": "logits"
+    }
+  ]
+}