From e6eaa7dcbc569f91f7b944c3e557680edc87fc8a Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Tue, 26 Nov 2024 01:07:49 +0000
Subject: [PATCH 01/12] update install to nightly releases

---
 .github/workflows/ci_eval.yaml       | 29 ++++------------------------
 .github/workflows/ci_eval_short.yaml | 18 ++---------------
 2 files changed, 6 insertions(+), 41 deletions(-)

diff --git a/.github/workflows/ci_eval.yaml b/.github/workflows/ci_eval.yaml
index 0164b6cdc..338b68553 100644
--- a/.github/workflows/ci_eval.yaml
+++ b/.github/workflows/ci_eval.yaml
@@ -57,22 +57,8 @@ jobs:
       - name: Install sharktank deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          # Note: We install in three steps in order to satisfy requirements
-          # from non default locations first. Installing the PyTorch CPU
-          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
-          pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
-
-          # Install latest iree-tubrine.
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
-
-          # Try with the latest IREE nightly releases, not what iree-turbine pins.
-          # We could also pin to a known working or stable version.
-          # This should eventually stabilize. Do the best we can for now.
-          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
-            iree-base-compiler \
-            iree-base-runtime
+          pip install shark-ai[apps]
+          python -m pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
 
       - name: Run perplexity test with IREE
         run:  pytest -n 8 -v -s sharktank/tests/evaluate/perplexity_iree_test.py --run-nightly-llama-tests --bs=100 --iree-device='hip://7' --iree-hip-target=gfx942 --iree-hal-target-backends=rocm --llama3-8b-f16-model-path=/data/llama3.1/8b/llama8b_f16.irpa --llama3-8b-tokenizer-path=/data/llama3.1/8b/tokenizer_config.json --html=out/llm/llama/perplexity/iree_perplexity/index.html
@@ -121,15 +107,8 @@ jobs:
       - name: Install sharktank deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          # Note: We install in three steps in order to satisfy requirements
-          # from non default locations first. Installing the PyTorch CPU
-          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
-          pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
-
-          # Install latest iree-tubrine.
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
+          pip install shark-ai[apps]
+          python -m pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
 
       - name: Run perplexity test with Torch
         run:  pytest -n 8 -v -s sharktank/tests/evaluate/perplexity_torch_test.py --longrun --llama3-8b-f16-model-path=/data/llama3.1/8b/llama8b_f16.irpa --llama3-8b-tokenizer-path=/data/llama3.1/8b/tokenizer_config.json --html=out/llm/llama/perplexity/torch_perplexity/index.html
diff --git a/.github/workflows/ci_eval_short.yaml b/.github/workflows/ci_eval_short.yaml
index 4622f5c57..830eed514 100644
--- a/.github/workflows/ci_eval_short.yaml
+++ b/.github/workflows/ci_eval_short.yaml
@@ -56,22 +56,8 @@ jobs:
       - name: Install sharktank deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          # Note: We install in three steps in order to satisfy requirements
-          # from non default locations first. Installing the PyTorch CPU
-          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
-          pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
-
-          # Install latest iree-tubrine.
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
-
-          # Try with the latest IREE nightly releases, not what iree-turbine pins.
-          # We could also pin to a known working or stable version.
-          # This should eventually stabilize. Do the best we can for now.
-          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
-            iree-base-compiler \
-            iree-base-runtime
+          pip install shark-ai[apps]
+          python -m pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
 
       - name: Run perplexity test with vmfb
         run:  pytest -n 8 -v -s sharktank/tests/evaluate/perplexity_iree_test.py --bs=5 --iree-device='hip://6' --iree-hip-target=gfx942 --iree-hal-target-backends=rocm --llama3-8b-f16-model-path=/data/llama3.1/8b/llama8b_f16.irpa --llama3-8b-tokenizer-path=/data/llama3.1/8b/tokenizer_config.json

From 4604c2b3ebdff520c2693fd64773fe3e7e49caae Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Tue, 26 Nov 2024 01:13:05 +0000
Subject: [PATCH 02/12] Test CI

---
 .github/workflows/ci_eval.yaml | 1 +
 1 file changed, 1 insertion(+)

diff --git a/.github/workflows/ci_eval.yaml b/.github/workflows/ci_eval.yaml
index 338b68553..8632ffbbe 100644
--- a/.github/workflows/ci_eval.yaml
+++ b/.github/workflows/ci_eval.yaml
@@ -7,6 +7,7 @@
 name: CI - sharktank perplexity
 
 on:
+  pull_request:
   workflow_dispatch:
   schedule:
     # Weekdays nightly at 07:00 UTC = 23:00 PST / 00:00 PDT.

From 7fdeaa1dd2fa8680b5865d5df7eb6bac0af796f3 Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Tue, 26 Nov 2024 04:55:48 +0000
Subject: [PATCH 03/12] Revert Test CI changes

---
 .github/workflows/ci_eval.yaml | 1 -
 1 file changed, 1 deletion(-)

diff --git a/.github/workflows/ci_eval.yaml b/.github/workflows/ci_eval.yaml
index 8632ffbbe..338b68553 100644
--- a/.github/workflows/ci_eval.yaml
+++ b/.github/workflows/ci_eval.yaml
@@ -7,7 +7,6 @@
 name: CI - sharktank perplexity
 
 on:
-  pull_request:
   workflow_dispatch:
   schedule:
     # Weekdays nightly at 07:00 UTC = 23:00 PST / 00:00 PDT.

From 1e5e4b981efb26bf62cbdd572f6b26d3d2877019 Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Tue, 26 Nov 2024 18:18:15 +0000
Subject: [PATCH 04/12] Update sharktank shortfin installation for CIs

---
 .github/workflows/ci-llama-large-tests.yaml   | 18 +++--------
 .github/workflows/ci-llama-quick-tests.yaml   | 17 ++---------
 .github/workflows/ci-sglang-benchmark.yml     | 19 ++++--------
 .../workflows/ci-sglang-integration-tests.yml | 18 ++++-------
 .github/workflows/ci-shark-ai.yml             | 19 ++----------
 .github/workflows/ci-sharktank.yml            | 30 +++----------------
 .github/workflows/ci_eval.yaml                | 10 +++++--
 .github/workflows/ci_eval_short.yaml          |  2 +-
 8 files changed, 32 insertions(+), 101 deletions(-)

diff --git a/.github/workflows/ci-llama-large-tests.yaml b/.github/workflows/ci-llama-large-tests.yaml
index 644066094..7765324cb 100644
--- a/.github/workflows/ci-llama-large-tests.yaml
+++ b/.github/workflows/ci-llama-large-tests.yaml
@@ -59,21 +59,11 @@ jobs:
       - name: Install pip deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          # Note: We install in three steps in order to satisfy requirements
-          # from non default locations first. Installing the PyTorch CPU
-          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
-          pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
-
-          # Install latest iree-turbine.
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
-
-
-          # Test with nightly releases, not what iree-turbine uses.
+          pip install shark-ai[apps]
+          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels --upgrade --pre
           pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
-            iree-base-compiler \
-            iree-base-runtime
+            iree-base-compiler iree-base-runtime --src deps \
+            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
 
       - name: Run llama tests
         run: pytest sharktank/tests/models/llama/benchmark_amdgpu_test.py -v -s --run-nightly-llama-tests --iree-hip-target=gfx942 --html=out/llm/llama/benchmark/index.html
diff --git a/.github/workflows/ci-llama-quick-tests.yaml b/.github/workflows/ci-llama-quick-tests.yaml
index a8c315ec8..d0019648d 100644
--- a/.github/workflows/ci-llama-quick-tests.yaml
+++ b/.github/workflows/ci-llama-quick-tests.yaml
@@ -59,21 +59,8 @@ jobs:
       - name: Install pip deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          # Note: We install in three steps in order to satisfy requirements
-          # from non default locations first. Installing the PyTorch CPU
-          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
-          pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
-
-          # Install latest iree-turbine.
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
-
-
-          # Test with nightly releases, not what iree-turbine uses.
-          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
-            iree-base-compiler \
-            iree-base-runtime
+          pip install shark-ai[apps]
+          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
 
       - name: Run llama 8b f16 decomposed test
         run: pytest sharktank/tests/models/llama/benchmark_amdgpu_test.py -v -s --iree-hip-target=gfx942 --run-quick-llama-test
diff --git a/.github/workflows/ci-sglang-benchmark.yml b/.github/workflows/ci-sglang-benchmark.yml
index f44e2772b..4ede69137 100644
--- a/.github/workflows/ci-sglang-benchmark.yml
+++ b/.github/workflows/ci-sglang-benchmark.yml
@@ -58,21 +58,12 @@ jobs:
       - name: Install pip deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          # Note: We install in three steps in order to satisfy requirements
-          # from non default locations first. Installing the PyTorch CPU
-          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
-          pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
+          pip install shark-ai[apps]
+          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels --upgrade --pre
+          pip install shortfin -f https://github.com/nod-ai/shark-ai/releases/expanded_assets/dev-wheels --upgrade --pre
+          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
+            iree-base-compiler iree-base-runtime --src deps \
             -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
-          pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
-
-          # Try with the latest nightly releases, not what iree-turbine pins.
-          # We could also pin to a known working or stable version.
-          # This should eventually stabilize. Do the best we can for now.
-          pip install -f https://iree.dev/pip-release-links.html --upgrade \
-            iree-base-compiler==3.0.0rc20241118 \
-            iree-base-runtime==3.0.0rc20241118 \
-            "numpy<2.0"
 
       - name: Install SGLang
         run: pip install "git+https://github.com/nod-ai/sglang.git#subdirectory=python"
diff --git a/.github/workflows/ci-sglang-integration-tests.yml b/.github/workflows/ci-sglang-integration-tests.yml
index c61756d78..398348cb2 100644
--- a/.github/workflows/ci-sglang-integration-tests.yml
+++ b/.github/workflows/ci-sglang-integration-tests.yml
@@ -59,20 +59,12 @@ jobs:
       - name: Install pip deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          # Note: We install in three steps in order to satisfy requirements
-          # from non default locations first. Installing the PyTorch CPU
-          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
-          pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
+          pip install shark-ai[apps]
+          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels --upgrade --pre
+          pip install shortfin -f https://github.com/nod-ai/shark-ai/releases/expanded_assets/dev-wheels --upgrade --pre
+          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
+            iree-base-compiler iree-base-runtime --src deps \
             -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
-          pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
-
-          # Use newest possible releases to be able to track commits that may
-          # cause errors.
-          pip install -f https://iree.dev/pip-release-links.html --upgrade \
-            iree-base-compiler \
-            iree-base-runtime \
-            "numpy<2.0"
 
       - name: Install SGLang
         run: pip install "git+https://github.com/nod-ai/sglang.git#subdirectory=python"
diff --git a/.github/workflows/ci-shark-ai.yml b/.github/workflows/ci-shark-ai.yml
index fc85a76a7..34440bf60 100644
--- a/.github/workflows/ci-shark-ai.yml
+++ b/.github/workflows/ci-shark-ai.yml
@@ -54,22 +54,9 @@ jobs:
       - name: Install pip deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          # Note: We install in three steps in order to satisfy requirements
-          # from non default locations first. Installing the PyTorch CPU
-          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
-          pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
-
-          # Install latest iree-tubrine.
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
-
-          # Try with the latest IREE nightly releases, not what iree-turbine pins.
-          # We could also pin to a known working or stable version.
-          # This should eventually stabilize. Do the best we can for now.
-          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
-            iree-base-compiler \
-            iree-base-runtime
+          pip install shark-ai[apps]
+          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
+          pip install shortfin -f https://github.com/nod-ai/shark-ai/releases/expanded_assets/dev-wheels
 
       - name: Run LLM Integration Tests
         run: pytest -v app_tests/integration_tests/llm/shortfin --log-cli-level=INFO
diff --git a/.github/workflows/ci-sharktank.yml b/.github/workflows/ci-sharktank.yml
index 1d3960b43..243a8b8ae 100644
--- a/.github/workflows/ci-sharktank.yml
+++ b/.github/workflows/ci-sharktank.yml
@@ -55,16 +55,8 @@ jobs:
       - name: Install pip deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          # Note: We install in three steps in order to satisfy requirements
-          # from non default locations first. Installing the PyTorch CPU
-          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
-          pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
-
-          # Update to the latest iree packages.
-          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
-            iree-base-compiler iree-base-runtime --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
+          pip install shark-ai[apps]
+          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
 
       - name: Run sharktank tests
         if: ${{ !cancelled() }}
@@ -107,22 +99,8 @@ jobs:
       - name: Install sharktank deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          # Note: We install in three steps in order to satisfy requirements
-          # from non default locations first. Installing the PyTorch CPU
-          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
-          pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
-
-          # Install latest iree-tubrine.
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
-
-          # Try with the latest IREE nightly releases, not what iree-turbine pins.
-          # We could also pin to a known working or stable version.
-          # This should eventually stabilize. Do the best we can for now.
-          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
-            iree-base-compiler \
-            iree-base-runtime
+          pip install shark-ai[apps]
+          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
 
       - name: Run tests
         run: |
diff --git a/.github/workflows/ci_eval.yaml b/.github/workflows/ci_eval.yaml
index 338b68553..39c8fcce9 100644
--- a/.github/workflows/ci_eval.yaml
+++ b/.github/workflows/ci_eval.yaml
@@ -58,7 +58,10 @@ jobs:
         run: |
           python -m pip install --no-compile --upgrade pip
           pip install shark-ai[apps]
-          python -m pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
+          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels --upgrade --pre
+          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
+            iree-base-compiler iree-base-runtime --src deps \
+            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
 
       - name: Run perplexity test with IREE
         run:  pytest -n 8 -v -s sharktank/tests/evaluate/perplexity_iree_test.py --run-nightly-llama-tests --bs=100 --iree-device='hip://7' --iree-hip-target=gfx942 --iree-hal-target-backends=rocm --llama3-8b-f16-model-path=/data/llama3.1/8b/llama8b_f16.irpa --llama3-8b-tokenizer-path=/data/llama3.1/8b/tokenizer_config.json --html=out/llm/llama/perplexity/iree_perplexity/index.html
@@ -108,7 +111,10 @@ jobs:
         run: |
           python -m pip install --no-compile --upgrade pip
           pip install shark-ai[apps]
-          python -m pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
+          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels --upgrade --pre
+          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
+            iree-base-compiler iree-base-runtime --src deps \
+            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
 
       - name: Run perplexity test with Torch
         run:  pytest -n 8 -v -s sharktank/tests/evaluate/perplexity_torch_test.py --longrun --llama3-8b-f16-model-path=/data/llama3.1/8b/llama8b_f16.irpa --llama3-8b-tokenizer-path=/data/llama3.1/8b/tokenizer_config.json --html=out/llm/llama/perplexity/torch_perplexity/index.html
diff --git a/.github/workflows/ci_eval_short.yaml b/.github/workflows/ci_eval_short.yaml
index 830eed514..7bb6a43c0 100644
--- a/.github/workflows/ci_eval_short.yaml
+++ b/.github/workflows/ci_eval_short.yaml
@@ -57,7 +57,7 @@ jobs:
         run: |
           python -m pip install --no-compile --upgrade pip
           pip install shark-ai[apps]
-          python -m pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
+          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
 
       - name: Run perplexity test with vmfb
         run:  pytest -n 8 -v -s sharktank/tests/evaluate/perplexity_iree_test.py --bs=5 --iree-device='hip://6' --iree-hip-target=gfx942 --iree-hal-target-backends=rocm --llama3-8b-f16-model-path=/data/llama3.1/8b/llama8b_f16.irpa --llama3-8b-tokenizer-path=/data/llama3.1/8b/tokenizer_config.json

From 5e479f32bbf28db91e9dc8822b01234ac803fb97 Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Tue, 26 Nov 2024 19:47:17 +0000
Subject: [PATCH 05/12] Test iree nightly

---
 .github/workflows/ci-llama-quick-tests.yaml | 17 +++++++++++++++--
 .github/workflows/ci_eval_short.yaml        | 12 ++++++++++--
 2 files changed, 25 insertions(+), 4 deletions(-)

diff --git a/.github/workflows/ci-llama-quick-tests.yaml b/.github/workflows/ci-llama-quick-tests.yaml
index d0019648d..a8c315ec8 100644
--- a/.github/workflows/ci-llama-quick-tests.yaml
+++ b/.github/workflows/ci-llama-quick-tests.yaml
@@ -59,8 +59,21 @@ jobs:
       - name: Install pip deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          pip install shark-ai[apps]
-          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
+          # Note: We install in three steps in order to satisfy requirements
+          # from non default locations first. Installing the PyTorch CPU
+          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
+          pip install --no-compile -r pytorch-cpu-requirements.txt
+          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
+
+          # Install latest iree-turbine.
+          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
+            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
+
+
+          # Test with nightly releases, not what iree-turbine uses.
+          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
+            iree-base-compiler \
+            iree-base-runtime
 
       - name: Run llama 8b f16 decomposed test
         run: pytest sharktank/tests/models/llama/benchmark_amdgpu_test.py -v -s --iree-hip-target=gfx942 --run-quick-llama-test
diff --git a/.github/workflows/ci_eval_short.yaml b/.github/workflows/ci_eval_short.yaml
index 7bb6a43c0..e1489819b 100644
--- a/.github/workflows/ci_eval_short.yaml
+++ b/.github/workflows/ci_eval_short.yaml
@@ -56,8 +56,16 @@ jobs:
       - name: Install sharktank deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          pip install shark-ai[apps]
-          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
+          # Note: We install in three steps in order to satisfy requirements
+          # from non default locations first. Installing the PyTorch CPU
+          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
+          pip install --no-compile -r pytorch-cpu-requirements.txt
+          pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
+
+          # Update to the latest iree packages.
+          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
+            iree-base-compiler iree-base-runtime --src deps \
+            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
 
       - name: Run perplexity test with vmfb
         run:  pytest -n 8 -v -s sharktank/tests/evaluate/perplexity_iree_test.py --bs=5 --iree-device='hip://6' --iree-hip-target=gfx942 --iree-hal-target-backends=rocm --llama3-8b-f16-model-path=/data/llama3.1/8b/llama8b_f16.irpa --llama3-8b-tokenizer-path=/data/llama3.1/8b/tokenizer_config.json

From e69fd1c43c4d033a9e5854d8aa46c955659c6101 Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Wed, 27 Nov 2024 00:26:03 +0000
Subject: [PATCH 06/12] Update stable/nightly install

---
 .github/workflows/ci-llama-large-tests.yaml   |  1 -
 .github/workflows/ci-llama-quick-tests.yaml   | 13 +++---------
 .github/workflows/ci-sglang-benchmark.yml     |  2 --
 .../workflows/ci-sglang-integration-tests.yml | 10 +++++++---
 .github/workflows/ci-shark-ai.yml             | 11 +++++++---
 .github/workflows/ci-sharktank.yml            | 20 +++++++++++++++----
 .github/workflows/ci_eval.yaml                |  2 --
 .github/workflows/ci_eval_short.yaml          |  6 ++----
 8 files changed, 36 insertions(+), 29 deletions(-)

diff --git a/.github/workflows/ci-llama-large-tests.yaml b/.github/workflows/ci-llama-large-tests.yaml
index 7765324cb..3d590a362 100644
--- a/.github/workflows/ci-llama-large-tests.yaml
+++ b/.github/workflows/ci-llama-large-tests.yaml
@@ -59,7 +59,6 @@ jobs:
       - name: Install pip deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          pip install shark-ai[apps]
           pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels --upgrade --pre
           pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
             iree-base-compiler iree-base-runtime --src deps \
diff --git a/.github/workflows/ci-llama-quick-tests.yaml b/.github/workflows/ci-llama-quick-tests.yaml
index a8c315ec8..1dbda87f3 100644
--- a/.github/workflows/ci-llama-quick-tests.yaml
+++ b/.github/workflows/ci-llama-quick-tests.yaml
@@ -63,17 +63,10 @@ jobs:
           # from non default locations first. Installing the PyTorch CPU
           # wheels saves multiple minutes and a lot of bandwidth on runner setup.
           pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
+          pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
 
-          # Install latest iree-turbine.
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
-
-
-          # Test with nightly releases, not what iree-turbine uses.
-          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
-            iree-base-compiler \
-            iree-base-runtime
+          # Get latest stable IREE release
+          pip install iree-turbine
 
       - name: Run llama 8b f16 decomposed test
         run: pytest sharktank/tests/models/llama/benchmark_amdgpu_test.py -v -s --iree-hip-target=gfx942 --run-quick-llama-test
diff --git a/.github/workflows/ci-sglang-benchmark.yml b/.github/workflows/ci-sglang-benchmark.yml
index 4ede69137..ce5116e9e 100644
--- a/.github/workflows/ci-sglang-benchmark.yml
+++ b/.github/workflows/ci-sglang-benchmark.yml
@@ -57,8 +57,6 @@ jobs:
 
       - name: Install pip deps
         run: |
-          python -m pip install --no-compile --upgrade pip
-          pip install shark-ai[apps]
           pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels --upgrade --pre
           pip install shortfin -f https://github.com/nod-ai/shark-ai/releases/expanded_assets/dev-wheels --upgrade --pre
           pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
diff --git a/.github/workflows/ci-sglang-integration-tests.yml b/.github/workflows/ci-sglang-integration-tests.yml
index 398348cb2..eb1932157 100644
--- a/.github/workflows/ci-sglang-integration-tests.yml
+++ b/.github/workflows/ci-sglang-integration-tests.yml
@@ -59,9 +59,13 @@ jobs:
       - name: Install pip deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          pip install shark-ai[apps]
-          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels --upgrade --pre
-          pip install shortfin -f https://github.com/nod-ai/shark-ai/releases/expanded_assets/dev-wheels --upgrade --pre
+          # Note: We install in three steps in order to satisfy requirements
+          # from non default locations first. Installing the PyTorch CPU
+          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
+          pip install --no-compile -r pytorch-cpu-requirements.txt
+          pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
+
+          # Update to the latest iree packages.
           pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
             iree-base-compiler iree-base-runtime --src deps \
             -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
diff --git a/.github/workflows/ci-shark-ai.yml b/.github/workflows/ci-shark-ai.yml
index 34440bf60..5f5b177c3 100644
--- a/.github/workflows/ci-shark-ai.yml
+++ b/.github/workflows/ci-shark-ai.yml
@@ -54,9 +54,14 @@ jobs:
       - name: Install pip deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          pip install shark-ai[apps]
-          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
-          pip install shortfin -f https://github.com/nod-ai/shark-ai/releases/expanded_assets/dev-wheels
+          # Note: We install in three steps in order to satisfy requirements
+          # from non default locations first. Installing the PyTorch CPU
+          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
+          pip install --no-compile -r pytorch-cpu-requirements.txt
+          pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
+
+          # Get latest stable IREE release
+          pip install iree-turbine
 
       - name: Run LLM Integration Tests
         run: pytest -v app_tests/integration_tests/llm/shortfin --log-cli-level=INFO
diff --git a/.github/workflows/ci-sharktank.yml b/.github/workflows/ci-sharktank.yml
index 243a8b8ae..806e22dba 100644
--- a/.github/workflows/ci-sharktank.yml
+++ b/.github/workflows/ci-sharktank.yml
@@ -55,8 +55,14 @@ jobs:
       - name: Install pip deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          pip install shark-ai[apps]
-          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
+          # Note: We install in three steps in order to satisfy requirements
+          # from non default locations first. Installing the PyTorch CPU
+          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
+          pip install --no-compile -r pytorch-cpu-requirements.txt
+          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
+
+          # Get latest stable IREE release
+          pip install iree-turbine
 
       - name: Run sharktank tests
         if: ${{ !cancelled() }}
@@ -99,8 +105,14 @@ jobs:
       - name: Install sharktank deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          pip install shark-ai[apps]
-          pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
+          # Note: We install in three steps in order to satisfy requirements
+          # from non default locations first. Installing the PyTorch CPU
+          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
+          pip install --no-compile -r pytorch-cpu-requirements.txt
+          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
+
+          # Get latest stable IREE release
+          pip install iree-turbine
 
       - name: Run tests
         run: |
diff --git a/.github/workflows/ci_eval.yaml b/.github/workflows/ci_eval.yaml
index 39c8fcce9..ea0531c05 100644
--- a/.github/workflows/ci_eval.yaml
+++ b/.github/workflows/ci_eval.yaml
@@ -57,7 +57,6 @@ jobs:
       - name: Install sharktank deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          pip install shark-ai[apps]
           pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels --upgrade --pre
           pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
             iree-base-compiler iree-base-runtime --src deps \
@@ -110,7 +109,6 @@ jobs:
       - name: Install sharktank deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          pip install shark-ai[apps]
           pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels --upgrade --pre
           pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
             iree-base-compiler iree-base-runtime --src deps \
diff --git a/.github/workflows/ci_eval_short.yaml b/.github/workflows/ci_eval_short.yaml
index e1489819b..7296cc0a8 100644
--- a/.github/workflows/ci_eval_short.yaml
+++ b/.github/workflows/ci_eval_short.yaml
@@ -62,10 +62,8 @@ jobs:
           pip install --no-compile -r pytorch-cpu-requirements.txt
           pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
 
-          # Update to the latest iree packages.
-          pip install -f https://iree.dev/pip-release-links.html --upgrade --pre \
-            iree-base-compiler iree-base-runtime --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
+          # Get latest stable IREE release
+          pip install iree-turbine
 
       - name: Run perplexity test with vmfb
         run:  pytest -n 8 -v -s sharktank/tests/evaluate/perplexity_iree_test.py --bs=5 --iree-device='hip://6' --iree-hip-target=gfx942 --iree-hal-target-backends=rocm --llama3-8b-f16-model-path=/data/llama3.1/8b/llama8b_f16.irpa --llama3-8b-tokenizer-path=/data/llama3.1/8b/tokenizer_config.json

From c7eca7136507c6f75a91188f16bbe0ddf6f6eb80 Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Wed, 27 Nov 2024 00:30:33 +0000
Subject: [PATCH 07/12] Test CI

---
 .github/workflows/ci-llama-large-tests.yaml | 1 +
 .github/workflows/ci_eval.yaml              | 1 +
 2 files changed, 2 insertions(+)

diff --git a/.github/workflows/ci-llama-large-tests.yaml b/.github/workflows/ci-llama-large-tests.yaml
index 3d590a362..9dd19cb70 100644
--- a/.github/workflows/ci-llama-large-tests.yaml
+++ b/.github/workflows/ci-llama-large-tests.yaml
@@ -7,6 +7,7 @@
 name: Llama Benchmarking Tests
 
 on:
+  pull_request:
   workflow_dispatch:
   schedule:
     # Weekdays at 4:00 AM UTC = 9:00 PM PST.
diff --git a/.github/workflows/ci_eval.yaml b/.github/workflows/ci_eval.yaml
index ea0531c05..d1d232d79 100644
--- a/.github/workflows/ci_eval.yaml
+++ b/.github/workflows/ci_eval.yaml
@@ -7,6 +7,7 @@
 name: CI - sharktank perplexity
 
 on:
+  pull_request:
   workflow_dispatch:
   schedule:
     # Weekdays nightly at 07:00 UTC = 23:00 PST / 00:00 PDT.

From 3f1fe1149daca26367d5ac06b8b298c9867d24ec Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Wed, 27 Nov 2024 02:59:35 +0000
Subject: [PATCH 08/12] Update stable/nightly install

---
 .github/workflows/ci-llama-quick-tests.yaml | 3 ++-
 .github/workflows/ci-shark-ai.yml           | 3 ++-
 .github/workflows/ci-sharktank.yml          | 6 ++++--
 .github/workflows/ci_eval_short.yaml        | 3 ++-
 4 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/.github/workflows/ci-llama-quick-tests.yaml b/.github/workflows/ci-llama-quick-tests.yaml
index 1dbda87f3..4ee8d24ed 100644
--- a/.github/workflows/ci-llama-quick-tests.yaml
+++ b/.github/workflows/ci-llama-quick-tests.yaml
@@ -66,7 +66,8 @@ jobs:
           pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
 
           # Get latest stable IREE release
-          pip install iree-turbine
+          pip install --src deps \
+            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
 
       - name: Run llama 8b f16 decomposed test
         run: pytest sharktank/tests/models/llama/benchmark_amdgpu_test.py -v -s --iree-hip-target=gfx942 --run-quick-llama-test
diff --git a/.github/workflows/ci-shark-ai.yml b/.github/workflows/ci-shark-ai.yml
index 5f5b177c3..16705712b 100644
--- a/.github/workflows/ci-shark-ai.yml
+++ b/.github/workflows/ci-shark-ai.yml
@@ -61,7 +61,8 @@ jobs:
           pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
 
           # Get latest stable IREE release
-          pip install iree-turbine
+          pip install --src deps \
+            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
 
       - name: Run LLM Integration Tests
         run: pytest -v app_tests/integration_tests/llm/shortfin --log-cli-level=INFO
diff --git a/.github/workflows/ci-sharktank.yml b/.github/workflows/ci-sharktank.yml
index 806e22dba..28d2ff7f7 100644
--- a/.github/workflows/ci-sharktank.yml
+++ b/.github/workflows/ci-sharktank.yml
@@ -62,7 +62,8 @@ jobs:
           pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
 
           # Get latest stable IREE release
-          pip install iree-turbine
+          pip install --src deps \
+            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
 
       - name: Run sharktank tests
         if: ${{ !cancelled() }}
@@ -112,7 +113,8 @@ jobs:
           pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
 
           # Get latest stable IREE release
-          pip install iree-turbine
+          pip install --src deps \
+            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
 
       - name: Run tests
         run: |
diff --git a/.github/workflows/ci_eval_short.yaml b/.github/workflows/ci_eval_short.yaml
index 7296cc0a8..821eb40cc 100644
--- a/.github/workflows/ci_eval_short.yaml
+++ b/.github/workflows/ci_eval_short.yaml
@@ -63,7 +63,8 @@ jobs:
           pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
 
           # Get latest stable IREE release
-          pip install iree-turbine
+          pip install --src deps \
+            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
 
       - name: Run perplexity test with vmfb
         run:  pytest -n 8 -v -s sharktank/tests/evaluate/perplexity_iree_test.py --bs=5 --iree-device='hip://6' --iree-hip-target=gfx942 --iree-hal-target-backends=rocm --llama3-8b-f16-model-path=/data/llama3.1/8b/llama8b_f16.irpa --llama3-8b-tokenizer-path=/data/llama3.1/8b/tokenizer_config.json

From 692a1976f1b45c0f6192c24d2b1e679e1ad727a6 Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Thu, 28 Nov 2024 00:48:32 +0000
Subject: [PATCH 09/12] Revert "Update stable/nightly install"

This reverts commit 3f1fe1149daca26367d5ac06b8b298c9867d24ec.
---
 .github/workflows/ci-shark-ai.yml    |  4 ++--
 .github/workflows/ci-sharktank.yml   |  4 ++--
 .github/workflows/ci_eval_short.yaml | 10 ++++++++--
 3 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/.github/workflows/ci-shark-ai.yml b/.github/workflows/ci-shark-ai.yml
index fc85a76a7..3f8e13065 100644
--- a/.github/workflows/ci-shark-ai.yml
+++ b/.github/workflows/ci-shark-ai.yml
@@ -60,8 +60,8 @@ jobs:
           pip install --no-compile -r pytorch-cpu-requirements.txt
           pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
 
-          # Install latest iree-tubrine.
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
+          # Get latest stable IREE release
+          pip install --src deps \
             -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
 
           # Try with the latest IREE nightly releases, not what iree-turbine pins.
diff --git a/.github/workflows/ci-sharktank.yml b/.github/workflows/ci-sharktank.yml
index 1d3960b43..0f499aa22 100644
--- a/.github/workflows/ci-sharktank.yml
+++ b/.github/workflows/ci-sharktank.yml
@@ -113,8 +113,8 @@ jobs:
           pip install --no-compile -r pytorch-cpu-requirements.txt
           pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
 
-          # Install latest iree-tubrine.
-          pip install --no-compile -f https://iree.dev/pip-release-links.html --src deps \
+          # Get latest stable IREE release
+          pip install --src deps \
             -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
 
           # Try with the latest IREE nightly releases, not what iree-turbine pins.
diff --git a/.github/workflows/ci_eval_short.yaml b/.github/workflows/ci_eval_short.yaml
index 830eed514..80a74cacb 100644
--- a/.github/workflows/ci_eval_short.yaml
+++ b/.github/workflows/ci_eval_short.yaml
@@ -56,8 +56,14 @@ jobs:
       - name: Install sharktank deps
         run: |
           python -m pip install --no-compile --upgrade pip
-          pip install shark-ai[apps]
-          python -m pip install sharktank -f https://github.com/nod-ai/SHARK-Platform/releases/expanded_assets/dev-wheels
+          # Note: We install in three steps in order to satisfy requirements
+          # from non default locations first. Installing the PyTorch CPU
+          # wheels saves multiple minutes and a lot of bandwidth on runner setup.
+          pip install --no-compile -r pytorch-cpu-requirements.txt
+          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
+
+          # Get latest stable IREE release
+          pip install iree-turbine==3.0.0
 
       - name: Run perplexity test with vmfb
         run:  pytest -n 8 -v -s sharktank/tests/evaluate/perplexity_iree_test.py --bs=5 --iree-device='hip://6' --iree-hip-target=gfx942 --iree-hal-target-backends=rocm --llama3-8b-f16-model-path=/data/llama3.1/8b/llama8b_f16.irpa --llama3-8b-tokenizer-path=/data/llama3.1/8b/tokenizer_config.json

From 810af5dc1eb027ddddb7e5fe6c1e7101c192a4b5 Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Thu, 28 Nov 2024 01:33:45 +0000
Subject: [PATCH 10/12] Update IREE to pinned versions

---
 .github/workflows/ci-llama-quick-tests.yaml |  5 ++---
 .github/workflows/ci-shark-ai.yml           |  5 ++---
 .github/workflows/ci-sharktank.yml          | 10 ++++------
 .github/workflows/ci_eval_short.yaml        |  2 +-
 4 files changed, 9 insertions(+), 13 deletions(-)

diff --git a/.github/workflows/ci-llama-quick-tests.yaml b/.github/workflows/ci-llama-quick-tests.yaml
index 4ee8d24ed..81d5ac5e8 100644
--- a/.github/workflows/ci-llama-quick-tests.yaml
+++ b/.github/workflows/ci-llama-quick-tests.yaml
@@ -65,9 +65,8 @@ jobs:
           pip install --no-compile -r pytorch-cpu-requirements.txt
           pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
 
-          # Get latest stable IREE release
-          pip install --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
+          # Get latest working stable IREE release
+          pip install iree-turbine==3.0.0
 
       - name: Run llama 8b f16 decomposed test
         run: pytest sharktank/tests/models/llama/benchmark_amdgpu_test.py -v -s --iree-hip-target=gfx942 --run-quick-llama-test
diff --git a/.github/workflows/ci-shark-ai.yml b/.github/workflows/ci-shark-ai.yml
index 16705712b..9c65a28de 100644
--- a/.github/workflows/ci-shark-ai.yml
+++ b/.github/workflows/ci-shark-ai.yml
@@ -60,9 +60,8 @@ jobs:
           pip install --no-compile -r pytorch-cpu-requirements.txt
           pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
 
-          # Get latest stable IREE release
-          pip install --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
+          # Get latest working stable IREE release
+          pip install iree-turbine==3.0.0
 
       - name: Run LLM Integration Tests
         run: pytest -v app_tests/integration_tests/llm/shortfin --log-cli-level=INFO
diff --git a/.github/workflows/ci-sharktank.yml b/.github/workflows/ci-sharktank.yml
index eadefa65f..b7cf57753 100644
--- a/.github/workflows/ci-sharktank.yml
+++ b/.github/workflows/ci-sharktank.yml
@@ -100,9 +100,8 @@ jobs:
           pip install --no-compile -r pytorch-cpu-requirements.txt
           pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
 
-          # Get latest stable IREE release
-          pip install --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
+          # Get latest working stable IREE release
+          pip install iree-turbine==3.0.0
 
       - name: Run sharktank tests
         if: ${{ !cancelled() }}
@@ -151,9 +150,8 @@ jobs:
           pip install --no-compile -r pytorch-cpu-requirements.txt
           pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
 
-          # Get latest stable IREE release
-          pip install --src deps \
-            -e "git+https://github.com/iree-org/iree-turbine.git#egg=iree-turbine"
+          # Get latest working stable IREE release
+          pip install iree-turbine==3.0.0
 
       - name: Run tests
         run: |
diff --git a/.github/workflows/ci_eval_short.yaml b/.github/workflows/ci_eval_short.yaml
index 80a74cacb..de55777dc 100644
--- a/.github/workflows/ci_eval_short.yaml
+++ b/.github/workflows/ci_eval_short.yaml
@@ -62,7 +62,7 @@ jobs:
           pip install --no-compile -r pytorch-cpu-requirements.txt
           pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
 
-          # Get latest stable IREE release
+          # Get latest working stable IREE release
           pip install iree-turbine==3.0.0
 
       - name: Run perplexity test with vmfb

From 8e3936e1e381651b5a8d44332800db391f6244b4 Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Thu, 28 Nov 2024 01:53:10 +0000
Subject: [PATCH 11/12] Remove shortfin for sharktank tests

---
 .github/workflows/ci-llama-quick-tests.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/ci-llama-quick-tests.yaml b/.github/workflows/ci-llama-quick-tests.yaml
index 81d5ac5e8..4425a0c9b 100644
--- a/.github/workflows/ci-llama-quick-tests.yaml
+++ b/.github/workflows/ci-llama-quick-tests.yaml
@@ -63,7 +63,7 @@ jobs:
           # from non default locations first. Installing the PyTorch CPU
           # wheels saves multiple minutes and a lot of bandwidth on runner setup.
           pip install --no-compile -r pytorch-cpu-requirements.txt
-          pip install --no-compile -r requirements.txt -e sharktank/ shortfin/
+          pip install --no-compile -r requirements.txt -r sharktank/requirements-tests.txt -e sharktank/
 
           # Get latest working stable IREE release
           pip install iree-turbine==3.0.0

From edf2ac6370a22cdab7a35f68abbb2fcd8137d6c3 Mon Sep 17 00:00:00 2001
From: archana-ramalingam <archana.ramalingam@amd.com>
Date: Thu, 28 Nov 2024 01:53:55 +0000
Subject: [PATCH 12/12] Revert changes to test nightly

---
 .github/workflows/ci-llama-large-tests.yaml | 1 -
 .github/workflows/ci_eval.yaml              | 1 -
 2 files changed, 2 deletions(-)

diff --git a/.github/workflows/ci-llama-large-tests.yaml b/.github/workflows/ci-llama-large-tests.yaml
index 9dd19cb70..3d590a362 100644
--- a/.github/workflows/ci-llama-large-tests.yaml
+++ b/.github/workflows/ci-llama-large-tests.yaml
@@ -7,7 +7,6 @@
 name: Llama Benchmarking Tests
 
 on:
-  pull_request:
   workflow_dispatch:
   schedule:
     # Weekdays at 4:00 AM UTC = 9:00 PM PST.
diff --git a/.github/workflows/ci_eval.yaml b/.github/workflows/ci_eval.yaml
index d1d232d79..ea0531c05 100644
--- a/.github/workflows/ci_eval.yaml
+++ b/.github/workflows/ci_eval.yaml
@@ -7,7 +7,6 @@
 name: CI - sharktank perplexity
 
 on:
-  pull_request:
   workflow_dispatch:
   schedule:
     # Weekdays nightly at 07:00 UTC = 23:00 PST / 00:00 PDT.