Upload PowerServe/Llama-3.1-8B-PowerServe-QNN29-8G4

Browse files

This view is limited to 50 files because it contains too many changes. See raw diff

Files changed (50) hide show

.gitattributes +9 -0
README.md +3 -0
ggml/weights.gguf +3 -0
model.json +27 -0
qnn/config.json +91 -0
qnn/kv/layer_0_key_0.raw +0 -0
qnn/kv/layer_0_key_1.raw +0 -0
qnn/kv/layer_0_value_0.raw +0 -0
qnn/kv/layer_0_value_1.raw +0 -0
qnn/kv/layer_10_key_0.raw +0 -0
qnn/kv/layer_10_key_1.raw +0 -0
qnn/kv/layer_10_value_0.raw +0 -0
qnn/kv/layer_10_value_1.raw +0 -0
qnn/kv/layer_11_key_0.raw +0 -0
qnn/kv/layer_11_key_1.raw +0 -0
qnn/kv/layer_11_value_0.raw +0 -0
qnn/kv/layer_11_value_1.raw +0 -0
qnn/kv/layer_12_key_0.raw +0 -0
qnn/kv/layer_12_key_1.raw +0 -0
qnn/kv/layer_12_value_0.raw +0 -0
qnn/kv/layer_12_value_1.raw +0 -0
qnn/kv/layer_13_key_0.raw +0 -0
qnn/kv/layer_13_key_1.raw +0 -0
qnn/kv/layer_13_value_0.raw +0 -0
qnn/kv/layer_13_value_1.raw +0 -0
qnn/kv/layer_14_key_0.raw +0 -0
qnn/kv/layer_14_key_1.raw +0 -0
qnn/kv/layer_14_value_0.raw +0 -0
qnn/kv/layer_14_value_1.raw +0 -0
qnn/kv/layer_15_key_0.raw +0 -0
qnn/kv/layer_15_key_1.raw +0 -0
qnn/kv/layer_15_value_0.raw +0 -0
qnn/kv/layer_15_value_1.raw +0 -0
qnn/kv/layer_16_key_0.raw +0 -0
qnn/kv/layer_16_key_1.raw +0 -0
qnn/kv/layer_16_value_0.raw +0 -0
qnn/kv/layer_16_value_1.raw +0 -0
qnn/kv/layer_17_key_0.raw +0 -0
qnn/kv/layer_17_key_1.raw +0 -0
qnn/kv/layer_17_value_0.raw +0 -0
qnn/kv/layer_17_value_1.raw +0 -0
qnn/kv/layer_18_key_0.raw +0 -0
qnn/kv/layer_18_key_1.raw +0 -0
qnn/kv/layer_18_value_0.raw +0 -0
qnn/kv/layer_18_value_1.raw +0 -0
qnn/kv/layer_19_key_0.raw +0 -0
qnn/kv/layer_19_key_1.raw +0 -0
qnn/kv/layer_19_value_0.raw +0 -0
qnn/kv/layer_19_value_1.raw +0 -0
qnn/kv/layer_1_key_0.raw +0 -0

.gitattributes CHANGED Viewed

@@ -33,3 +33,12 @@ 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

 *.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
+qnn/lm_head.bin filter=lfs diff=lfs merge=lfs -text
+qnn/smallthinker_3b_1.bin filter=lfs diff=lfs merge=lfs -text
+qnn/smallthinker_3b_0.bin filter=lfs diff=lfs merge=lfs -text
+ggml/weights.gguf filter=lfs diff=lfs merge=lfs -text
+.git/lfs/objects/c7/35/c7357a12af21aa21f1145c82f3673a98e920d837a1ed362cd6c47179ff63a33d filter=lfs diff=lfs merge=lfs -text
+.git/lfs/objects/27/a3/27a392dcfb4b6871d49c5dcba926dd591ebefeb3f8081510f6ee5154f97d16a1 filter=lfs diff=lfs merge=lfs -text
+.git/lfs/objects/41/52/4152b2401a22f48df3d3983733a096094bd69551ac0eff10de8ffaf5c42c9241 filter=lfs diff=lfs merge=lfs -text
+.git/lfs/objects/70/98/709854441cf8454de2a0e939a4749468ef355518668a5e0e25932f46ff2c1708 filter=lfs diff=lfs merge=lfs -text

README.md ADDED Viewed

	@@ -0,0 +1,3 @@

+---
+license: apache-2.0
+---

ggml/weights.gguf ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:311af5e718008431b60b8dc4c3613e835f04d4168ddd97c167277bf795657927
+size 1917656096

model.json ADDED Viewed

	@@ -0,0 +1,27 @@

+{
+    "model_id": "smallthinker_0.5b_8gen4",
+    "model_arch": "qwen2",
+    "version": 20241202,
+    "llm_config": {
+        "embed_dim": 896,
+        "ffn_dim": 4864,
+        "head_size": 64,
+        "kv_dim": 128,
+        "n_attn_heads": 14,
+        "n_attn_kv_heads": 2,
+        "n_ctx": 32768,
+        "n_layers": 24,
+        "norm_eps": 9.999999974752427e-07,
+        "vocab_size": 151936,
+        "rope_config": {
+            "n_rope_ctx_orig": 32768,
+            "rope_attn_factor": 1.0,
+            "rope_dim": 64,
+            "rope_freq_base": 1000000.0,
+            "rope_freq_scale": 1.0,
+            "rope_scale_type": "linear",
+            "rope_type": -1
+        }
+    },
+    "vision": {}
+}

qnn/config.json ADDED Viewed

	@@ -0,0 +1,91 @@

+{
+  "model_parameters": {
+    "n_layers": 36,
+    "vocab_size": 151936,
+    "embed_dim": 2048,
+    "ffn_hidden_dim": 11008,
+    "head_dim": 128,
+    "n_kv_heads": 2,
+    "rope_theta": 1000000.0,
+    "rms_norm_eps": 1e-06,
+    "attention_mask_value": -50000.0,
+    "tie_embedding": true
+  },
+  "qnn_parameters": {
+    "n_hvx_threads": 4
+  },
+  "graphs": [
+    {
+      "type": "transformers",
+      "start_layer_id": 0,
+      "end_layer_id": 18,
+      "batch_size": 16,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_16",
+      "model_path": "smallthinker_3b_0.bin",
+      "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
+      "kv_size": 11,
+      "x_name": "x",
+      "out_name": "out"
+    },
+    {
+      "type": "transformers",
+      "start_layer_id": 18,
+      "end_layer_id": 36,
+      "batch_size": 16,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_16",
+      "model_path": "smallthinker_3b_1.bin",
+      "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
+      "kv_size": 11,
+      "x_name": "x",
+      "out_name": "out"
+    },
+    {
+      "type": "transformers",
+      "start_layer_id": 0,
+      "end_layer_id": 18,
+      "batch_size": 128,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_128",
+      "model_path": "smallthinker_3b_0.bin",
+      "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
+      "kv_size": 11,
+      "x_name": "x",
+      "out_name": "out"
+    },
+    {
+      "type": "transformers",
+      "start_layer_id": 18,
+      "end_layer_id": 36,
+      "batch_size": 128,
+      "cache_size": 1920,
+      "context_size": 2048,
+      "graph_name": "batch_128",
+      "model_path": "smallthinker_3b_1.bin",
+      "kv_path_format": "kv/layer_{layer_id}_{kv_type}_{head_id}.raw",
+      "kv_size": 11,
+      "x_name": "x",
+      "out_name": "out"
+    }
+  ],
+  "embeddings": [
+    {
+      "graph_name": "batch_16",
+      "model_path": "lm_head.bin",
+      "batch_size": 16,
+      "x_name": "x",
+      "out_name": "logits"
+    },
+    {
+      "graph_name": "batch_128",
+      "model_path": "lm_head.bin",
+      "batch_size": 128,
+      "x_name": "x",
+      "out_name": "logits"
+    }
+  ]
+}

qnn/kv/layer_0_key_0.raw ADDED Viewed