From 99182d61038045f94d3b41100a8e4b188c96f60c Mon Sep 17 00:00:00 2001 From: Raymond Kim Date: Thu, 30 May 2024 10:38:34 -0400 Subject: [PATCH] #8764: Add disclaimer that current WH models are only on N300 and fix installing link in docs --- README.md | 6 +++++- docs/source/tt-metalium/get_started/get_started.rst | 2 +- 2 files changed, 6 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 42c69e0764bc..86373595e3f1 100644 --- a/README.md +++ b/README.md @@ -38,6 +38,10 @@ ## Wormhole (WH) Models +> [!NOTE] +> +> We currently test our demo models for Wormhole on N300 boards only. + | Model | Gen. Token [3] | Batch | End-to-end throughput [1] | Device throughput [2] | Target | |-------------------------------------------------------------|--------------------|----------------------|------------------------------|-----------------------------|----------------| | [Falcon7B-decode](./models/demos/wormhole/falcon7b) | 129th | 32 | 11.6 t/s/u - 371 t/s | 15.4 t/s/u - 493 t/s | 21 t/s/u | @@ -46,7 +50,7 @@ | [BERT-Large](./models/demos/metal_BERT_large_11/) (sen/s) | any | 8 | 270 | 340 | 400 | | [Stable Diffusion 1.4](./models/demos/wormhole/stable_diffusion) 512x512 (sec/img) | | 1 | 8s | 5s | | -[3] - Generating the i'th token in a sequence while the kv_cache is filled with i-1 rows. +[3] - Generating the `i`'th token in a sequence while the kv_cache is filled with `i-1` rows. ## T3000 (2x4 mesh of WHs) Models diff --git a/docs/source/tt-metalium/get_started/get_started.rst b/docs/source/tt-metalium/get_started/get_started.rst index 155054fd885d..9b7f848c5896 100644 --- a/docs/source/tt-metalium/get_started/get_started.rst +++ b/docs/source/tt-metalium/get_started/get_started.rst @@ -26,7 +26,7 @@ hardware. Install tt-metal and build the project by following the instructions in the `installation guide -`_. +`_. 2. Explore the Falcon 7B Demo ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^