Model-References content for 1.20.0

Signed-off-by: SW publisher <[email protected]>

Author: SW publisher

MLPERF4.0/Inference/llama/README.md

@@ -52,7 +52,7 @@ docker run --privileged --security-opt seccomp=unconfined   \
            -v $INTEL_HABANALABS_DIR:/root/Intel-HabanaLabs/ \
            --cap-add=sys_nice --cap-add=SYS_PTRACE          \
            --user root --workdir=/root --net=host           \
-           --ulimit memlock=-1:-1 vault.habana.ai/gaudi-docker/1.19.0/ubuntu22.04/habanalabs/pytorch-installer-2.5.1:latest
+           --ulimit memlock=-1:-1 vault.habana.ai/gaudi-docker/1.20.0/ubuntu22.04/habanalabs/pytorch-installer-2.6.0:latest
 ```
 ```bash
 docker exec -it mlperf-intel-habanalabs bash
@@ -164,6 +164,6 @@ The Quantization Toolkit is described in the [Intel Gaudi documentation](https:/
 
 ## Supported Configurations
 
-| Validated on | Intel Gaudi Software Version | Framework Version(s) |   Mode   |
-| :----------: | :--------------------------: | :------------------: | :------: |
-|    Gaudi 2   |      1.19.0                   |    PyTorch 2.5.1     | Inference |
+| Validated on | Intel Gaudi Software Version | Framework Version(s) |   Mode    |
+| :----------: | :--------------------------: | :------------------: | :-------: |
+|    Gaudi 2   |      1.20.0                  |    PyTorch 2.6.0     | Inference |

MLPERF4.0/Inference/llama/configs/fp8.conf

@@ -1,4 +1,4 @@
 *.Offline.min_query_count = 98304
 
-*.Server.target_qps = 21.1
+*.Server.target_qps = 20.2
 *.Server.min_query_count = 24576

MLPERF4.0/Inference/llama/setup_tgi.sh

@@ -23,15 +23,16 @@ pip install pip --upgrade
 # don't try to overwrite torch
 grep -v "torch==" requirements.txt | pip install --no-deps -r /dev/stdin
 pip install -e .
+# this stopped to be installed by TGI but is still required:
+pip install outlines==0.0.36
 cd ..
 # build router
 cd router
 cargo install --locked --path .
 cd ..
 # build launcher
 cd launcher
-cargo install --locked --path .
+cargo install --locked --path . 
 cd ..
 popd
-
 pip list

MLPERF4.0/Inference/stable-diffusion-xl/README.md

@@ -131,9 +131,9 @@ All necessary environmental variables are enabled by default.
 
 ## Supported Configurations
 
-| Validated on | Intel Gaudi Software Version | Framework Version(s) |   Mode   |
-| :----------: | :--------------------------: | :------------------: | :------: |
-|    Gaudi 2   |      1.18.0                   |    PyTorch 2.4.0     | Inference |
+| Validated on | Intel Gaudi Software Version | Framework Version(s) |   Mode    |
+| :----------: | :--------------------------: | :------------------: | :-------: |
+|    Gaudi 2   |      1.20.0                  |    PyTorch 2.6.0     | Inference |
 
 ## Changelog
 

MLPERF4.0/Training/benchmarks/llm_finetune/README.md

@@ -71,7 +71,7 @@ NOTE: The below is only an example value. Replace [Intel_Gaudi_software_version]
 
 ### Download Data and Model
 
-MLCommons hosts the model for download exclusively by MLCommons Members. You must first agree to the [confidentiality notice](https://docs.google.com/forms/d/e/1FAIpQLSc_8VIvRmXM3I8KQaYnKf7gy27Z63BBoI_I1u02f4lw6rBp3g/viewform), then follow the [link[(https://drive.google.com/drive/folders/11tBZvvrh0FCm3XuR5E849K42TqftYdUF)] to a directory containing [Rclone download instructions](https://docs.google.com/document/d/1Yp2T_TsVfg8uEoEv0wa-dGP4R7r1EOHucTvDNWznWzE/edit#heading=h.at8a3matgbrk). Follow steps 1-3 to install and activate Rclone. Finally, download the model to the desired download directory (default ./models):
+MLCommons hosts the model for download exclusively by MLCommons Members. You must first agree to the [confidentiality notice](https://docs.google.com/forms/d/e/1FAIpQLSc_8VIvRmXM3I8KQaYnKf7gy27Z63BBoI_I1u02f4lw6rBp3g/viewform), then follow the [link](https://drive.google.com/drive/folders/11tBZvvrh0FCm3XuR5E849K42TqftYdUF) to a directory containing [Rclone download instructions](https://docs.google.com/document/d/1Yp2T_TsVfg8uEoEv0wa-dGP4R7r1EOHucTvDNWznWzE/edit#heading=h.at8a3matgbrk). Follow steps 1-3 to install and activate Rclone. Finally, download the model to the desired download directory (default ./models):
 Log into mlperf4.0 container and run:
 ```bash
 rclone copy mlc-llama2:Llama2-70b-fused-qkv-mlperf /root/model/Llama2-70b-fused-qkv-mlperf -P
@@ -107,4 +107,4 @@ cp /root/MLPERF/Intel-HabanaLabs/benchmarks/llm_finetune/config.json /root/model
 
 | Validated on | Intel Gaudi Software Version | Framework Version(s) |   Mode   |
 | :----------: | :--------------------------: | :------------------: | :------: |
-|    Gaudi 2   |      1.18.0                   |    PyTorch 2.4.0     | Training |
+|    Gaudi 2   |      1.20.0                  |    PyTorch 2.6.0     | Training |