From c5ca91e542fa6d67bddb92f6fdad0f8834b986a7 Mon Sep 17 00:00:00 2001 From: andrej Date: Tue, 24 Mar 2026 16:13:18 -0600 Subject: [PATCH 1/2] fix FLOP/s throughput metric in CI --- .github/actions/commit_results/action.yaml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/actions/commit_results/action.yaml b/.github/actions/commit_results/action.yaml index e5fa277b..00324f66 100644 --- a/.github/actions/commit_results/action.yaml +++ b/.github/actions/commit_results/action.yaml @@ -6,7 +6,7 @@ name: "Commit Test Results to a Branch" inputs: pretty_flags: required: false - default: "--metric='Latency (mean)' --metric='Bandwidth (mean)' --metric='GFLOP/s (mean)' -o pretty.md" + default: "--metric='Latency (mean)' --metric='Bandwidth (mean)' --metric='Throughput (mean)' -o pretty.md" input_csv: required: false default: "tests_latest.csv" From 6ec95fd691ea4ce4837ab8d5070f460dfb2e2611 Mon Sep 17 00:00:00 2001 From: andrej Date: Wed, 25 Mar 2026 13:25:05 -0600 Subject: [PATCH 2/2] also fix tokens per second output --- iron/applications/llama_3.2_1b/test.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/iron/applications/llama_3.2_1b/test.py b/iron/applications/llama_3.2_1b/test.py index 18f4c602..8683615e 100644 --- a/iron/applications/llama_3.2_1b/test.py +++ b/iron/applications/llama_3.2_1b/test.py @@ -32,7 +32,7 @@ def generate_test_params(): @pytest.mark.metrics( TTFT=r"\[Prefill\]\s*Time to first token:\s*(?P[\d\.e\+-]+) s", - TPS=r"\[Decode\]\s*Tokens per second: (?P[\d\.e\+-]+)", + TPS=r"\[Decode\]\s*Tokens per second:\s*(?P[\d\.e\+-]+)", ) @pytest.mark.parametrize("prompt_len,num_tokens", params, ids=names) def test_llama_3_2_1b(prompt_len, num_tokens):