Llamafile 1.2.0

pts/llamafile-1.2.0 - 02 June 2024 - Update against Llamafile 0.8.6 upstream.

downloads.xml

<?xml version="1.0"?>
<!--Phoronix Test Suite v10.8.5-->
<PhoronixTestSuite>
  <Downloads>
    <Package>
      <URL>https://huggingface.co/Mozilla/Mistral-7B-Instruct-v0.2-llamafile/resolve/772507f9981f04bca0b745027b9a9cf585cb490c/mistral-7b-instruct-v0.2.Q5_K_M.llamafile?download=true</URL>
      <MD5>991dd4c8a4ca30740854b9dc82286ae1</MD5>
      <SHA256>f68c1e37d4d2100b9acbf5ccf806461ab6c8e47c351fb69133eb8aa545b06883</SHA256>
      <FileName>mistral-7b-instruct-v0.2.Q5_K_M.llamafile.86</FileName>
      <FileSize>5166938210</FileSize>
      <Optional>TRUE</Optional>
    </Package>
    <Package>
      <URL>https://huggingface.co/Mozilla/llava-v1.5-7b-llamafile/resolve/16d41be6ec04c0eb5397bebf49c0736ad263df2c/llava-v1.5-7b-q4.llamafile?download=true</URL>
      <MD5>c88dde794dcbe349a12c88c93cd8cec7</MD5>
      <SHA256>044090e12df362c3ff72151dc22f6be086af694f3f64dc3db6f61f89256175b8</SHA256>
      <FileName>llava-v1.6-mistral-7b.Q8_0.llamafile.86</FileName>
      <FileSize>8064350537</FileSize>
      <Optional>TRUE</Optional>
    </Package>
    <Package>
      <URL>https://huggingface.co/Mozilla/WizardCoder-Python-34B-V1.0-llamafile/resolve/f7bf923573f2d667551d098bb36822571f2b6c11/wizardcoder-python-34b-v1.0.Q6_K.llamafile?download=true</URL>
      <MD5>4a07ecb839f534f95d7acc25f33f7f01</MD5>
      <SHA256>97480e6106fdc337ae23e0c7d6d842be260ffb35e0b76d75d3f02e5341294e5a</SHA256>
      <FileName>wizardcoder-python-34b-v1.0.Q6_K.llamafile.86</FileName>
      <FileSize>27708232181</FileSize>
      <Optional>TRUE</Optional>
    </Package>
    <Package>
      <URL>https://huggingface.co/Mozilla/Meta-Llama-3-8B-Instruct-llamafile/resolve/f8eb0593953dbfff22f9eb61a151be1147caf263/Meta-Llama-3-8B-Instruct.F16.llamafile?download=true</URL>
      <MD5>6191ce33888e6d48ef9b54b66e6d6b2f</MD5>
      <SHA256>5773b73e471ab1a4f31bde1fcf7ce18963427eca197a6658a260412bbe7b0dfe</SHA256>
      <FileName>Meta-Llama-3-8B-Instruct.F16.llamafile.86</FileName>
      <FileSize>16104503597</FileSize>
      <Optional>TRUE</Optional>
    </Package>
    <Package>
      <URL>https://huggingface.co/Mozilla/TinyLlama-1.1B-Chat-v1.0-llamafile/resolve/2c0dd78a401477ae7f9f8dd03786880aedfe07f0/TinyLlama-1.1B-Chat-v1.0.BF16.llamafile?download=true</URL>
      <MD5>3a10c549990cd3cdf0320b04cfdd1381</MD5>
      <SHA256>255395ca614b39877fa4daede7d7ecb989f3172d8058f76ddf5f24e7139b38c4</SHA256>
      <FileName>TinyLlama-1.1B-Chat-v1.0.BF16.llamafile.86</FileName>
      <FileSize>2236495966</FileSize>
      <Optional>TRUE</Optional>
    </Package>
    <Package>
      <URL>http://www.phoronix-test-suite.com/benchmark-files/pts-sample-photos-2.tar.bz2, http://www.phoronix.net/downloads/phoronix-test-suite/benchmark-files/pts-sample-photos-2.tar.bz2</URL>
      <MD5>7ff2870cf3fa2299ea5f4e77e5a8535f</MD5>
      <SHA256>e8597d7d5910a709985bffae490dfa9490483eb7838b351191156f027d9b5272</SHA256>
      <FileName>pts-sample-photos-2.tar.bz2</FileName>
      <FileSize>29599729</FileSize>
    </Package>
  </Downloads>
</PhoronixTestSuite>

install.sh

#!/bin/bash
chmod +x mistral-7b-instruct-v0.2.Q5_K_M.llamafile.86
chmod +x llava-v1.6-mistral-7b.Q8_0.llamafile.86
chmod +x wizardcoder-python-34b-v1.0.Q6_K.llamafile.86
chmod +x Meta-Llama-3-8B-Instruct.F16.llamafile.86
chmod +x TinyLlama-1.1B-Chat-v1.0.BF16.llamafile.86
tar -xf pts-sample-photos-2.tar.bz2
echo $? > ~/install-exit-status

cat <<'EOT' > run-mistral
#!/bin/bash
./mistral-7b-instruct-v0.2.Q5_K_M.llamafile.86 --temp 0.7 -p '[INST]Write a long story about llamas[/INST]' $@
exit $?
EOT
chmod +x run-mistral

cat <<'EOT' > run-llama3
#!/bin/bash
./Meta-Llama-3-8B-Instruct.F16.llamafile.86 --temp 0.7 -p '[INST]Write a long story about llamas[/INST]' $@
exit $?
EOT
chmod +x run-llama3

cat <<'EOT' > run-tinyllama
#!/bin/bash
./TinyLlama-1.1B-Chat-v1.0.BF16.llamafile.86 --temp 0.7 -p '[INST]Write a long story about llamas[/INST]' $@
exit $?
EOT
chmod +x run-tinyllama

cat <<'EOT' > run-wizardcoder
#!/bin/bash
./wizardcoder-python-34b-v1.0.Q6_K.llamafile.86 --temp 0 -e -r '```\n' -p '```c\nvoid *memcpy_sse2(char *dst, const char *src, size_t size) {\n' $@
exit $?
EOT
chmod +x run-wizardcoder

cat <<'EOT' > run-llava
#!/bin/bash
./llava-v1.6-mistral-7b.Q8_0.llamafile.86 --temp 0.2 --image DSC_4646.JPG -e -p '### User: Describe in detail what do you see?\n### Assistant:' $@
exit $?
EOT
chmod +x run-llava

echo "#!/bin/sh
./\$@ -t \$NUM_CPU_PHYSICAL_CORES > \$LOG_FILE 2>&1
echo \$? > ~/test-exit-status" > ~/llamafile
chmod +x ~/llamafile

results-definition.xml

<?xml version="1.0"?>
<!--Phoronix Test Suite v10.8.5-->
<PhoronixTestSuite>
  <ResultsParser>
    <OutputTemplate>llama_print_timings:        eval time =   18329.86 ms /   399 runs   (   45.94 ms per token,    #_RESULT_# tokens per second)</OutputTemplate>
    <LineHint>tokens per second</LineHint>
    <ResultBeforeString>tokens</ResultBeforeString>
  </ResultsParser>
</PhoronixTestSuite>

test-definition.xml

<?xml version="1.0"?>
<!--Phoronix Test Suite v10.8.5-->
<PhoronixTestSuite>
  <TestInformation>
    <Title>Llamafile</Title>
    <AppVersion>0.8.6</AppVersion>
    <Description>Mozilla's Llamafile allows distributing and running large language models (LLMs) as a single file. Llamafile aims to make open-source LLMs more accessible to developers and users. Llamafile supports a variety of models, CPUs and GPUs, and other options.</Description>
    <ResultScale>Tokens Per Second</ResultScale>
    <Proportion>HIB</Proportion>
    <TimesToRun>3</TimesToRun>
  </TestInformation>
  <TestProfile>
    <Version>1.2.0</Version>
    <SupportedPlatforms>Linux</SupportedPlatforms>
    <SoftwareType>Utility</SoftwareType>
    <TestType>System</TestType>
    <License>Free</License>
    <EnvironmentSize>40000</EnvironmentSize>
    <ProjectURL>https://llamafile.ai/</ProjectURL>
    <RepositoryURL>https://github.com/Mozilla-Ocho/llamafile</RepositoryURL>
    <Maintainer>Michael Larabel</Maintainer>
  </TestProfile>
  <TestSettings>
    <Option>
      <DisplayName>Test</DisplayName>
      <Identifier>test</Identifier>
      <Menu>
        <Entry>
          <Name>mistral-7b-instruct-v0.2.Q5_K_M</Name>
          <Value>run-mistral</Value>
        </Entry>
        <Entry>
          <Name>Meta-Llama-3-8B-Instruct.F16</Name>
          <Value>run-llama3</Value>
        </Entry>
        <Entry>
          <Name>TinyLlama-1.1B-Chat-v1.0.BF16</Name>
          <Value>run-tinyllama</Value>
        </Entry>
        <Entry>
          <Name>llava-v1.6-mistral-7b.Q8_0</Name>
          <Value>run-llava</Value>
        </Entry>
        <Entry>
          <Name>wizardcoder-python-34b-v1.0.Q6_K</Name>
          <Value>run-wizardcoder</Value>
        </Entry>
      </Menu>
    </Option>
    <Option>
      <DisplayName>Acceleration</DisplayName>
      <Identifier>accel</Identifier>
      <Menu>
        <Entry>
          <Name>CPU</Name>
          <Value>--gpu DISABLE</Value>
        </Entry>
        <Entry>
          <Name>GPU AUTO</Name>
          <Value>--gpu AUTO</Value>
        </Entry>
      </Menu>
    </Option>
  </TestSettings>
</PhoronixTestSuite>