From 0c4598d3e06883bf9d2cad9710c03ad6859c8306 Mon Sep 17 00:00:00 2001
From: Ahmed Ahmed <ahmedah@stanford.edu>
Date: Sun, 17 Nov 2024 19:50:42 -0800
Subject: [PATCH] llama 2 hf convert yaml

---
 llama2_hf_convert.yaml | 10 ++++++++++
 1 file changed, 10 insertions(+)
 create mode 100644 llama2_hf_convert.yaml

diff --git a/llama2_hf_convert.yaml b/llama2_hf_convert.yaml
new file mode 100644
index 000000000..4394c327c
--- /dev/null
+++ b/llama2_hf_convert.yaml
@@ -0,0 +1,10 @@
+model:  # 7B class model
+  type: llama
+  seq_len: 2048
+  hidden_dim: 4096
+  intermediate_dim: 11008
+  num_layers: 32
+  num_heads: 32
+  num_kv_heads: 32
+  use_flash_attention: True
+  flash_attention_block_size: 1024
\ No newline at end of file