RSR-27B / assets /sample_trajectories.svg
zli12321's picture
Add sample trajectories section and chart to model card
1ad6e66 verified
|
Raw History Blame Contribute Delete
5.97 kB
<svg xmlns="http://www.w3.org/2000/svg" width="960" height="432" viewBox="0 0 960 432" font-family="-apple-system, BlinkMacSystemFont, 'Segoe UI', Helvetica, Arial, sans-serif">
<rect x="0.5" y="0.5" width="959" height="431" rx="8" fill="#ffffff" stroke="#d1d9e0"/>
<text x="24" y="34" font-size="17" font-weight="600" fill="#1f2328">Expert trajectory vs. passing rewrite on 3 sample SWR tasks</text>
<text x="24" y="58" font-size="12.5" fill="#59636e">Bars show commands run, colored by kind; labels give steps, commands, and agent time.</text>
<text x="24" y="76" font-size="12.5" fill="#59636e">Expert: base Qwen3.8-27B under continue-until-timeout. Rewrite: base Qwen3.8-27B as plain Terminus-2, runbook kept private.</text>
<text x="24" y="114" font-size="13.5"><tspan font-weight="600" fill="#1f2328">tree_sitter_markdown_inline__05</tspan><tspan fill="#59636e" font-size="12.5"> 路 Data, query &amp; formal (tree-sitter)</tspan></text>
<text x="40" y="140.5" font-size="12.5" fill="#59636e">Expert</text>
<rect x="120" y="126" width="560" height="20" rx="3" fill="#f6f8fa"/>
<rect x="120.00" y="126" width="318.43" height="20" fill="#6e7781"><title>explore: 87</title></rect>
<rect x="438.43" y="126" width="58.56" height="20" fill="#9a6700"><title>edit: 16</title></rect>
<rect x="496.99" y="126" width="87.84" height="20" fill="#1a7f37"><title>run: 24</title></rect>
<rect x="584.84" y="126" width="7.32" height="20" fill="#afb8c1"><title>wait: 2</title></rect>
<rect x="592.16" y="126" width="87.84" height="20" fill="#57606a"><title>other: 24</title></rect>
<text x="690.00" y="140.5" font-size="12.5" fill="#1f2328">64 steps 路 153 commands 路 2.0 h</text>
<text x="40" y="168.5" font-size="12.5" fill="#59636e">Rewrite</text>
<rect x="120" y="154" width="560" height="20" rx="3" fill="#f6f8fa"/>
<rect x="120.00" y="154" width="120.78" height="20" fill="#6e7781"><title>explore: 33</title></rect>
<rect x="240.78" y="154" width="32.94" height="20" fill="#9a6700"><title>edit: 9</title></rect>
<rect x="273.73" y="154" width="91.50" height="20" fill="#1a7f37"><title>run: 25</title></rect>
<rect x="365.23" y="154" width="7.32" height="20" fill="#8250df"><title>install: 2</title></rect>
<rect x="372.55" y="154" width="32.94" height="20" fill="#57606a"><title>other: 9</title></rect>
<text x="415.49" y="168.5" font-size="12.5" fill="#1f2328">34 steps 路 78 commands 路 22 min</text>
<text x="24" y="210" font-size="13.5"><tspan font-weight="600" fill="#1f2328">cmake_generated_header_config__06</tspan><tspan fill="#59636e" font-size="12.5"> 路 Systems &amp; infrastructure (CMake)</tspan></text>
<text x="40" y="236.5" font-size="12.5" fill="#59636e">Expert</text>
<rect x="120" y="222" width="560" height="20" rx="3" fill="#f6f8fa"/>
<rect x="120.00" y="222" width="161.05" height="20" fill="#6e7781"><title>explore: 44</title></rect>
<rect x="281.05" y="222" width="21.96" height="20" fill="#9a6700"><title>edit: 6</title></rect>
<rect x="303.01" y="222" width="142.75" height="20" fill="#1a7f37"><title>run: 39</title></rect>
<rect x="445.75" y="222" width="18.30" height="20" fill="#57606a"><title>other: 5</title></rect>
<text x="474.05" y="236.5" font-size="12.5" fill="#1f2328">59 steps 路 94 commands 路 2.0 h</text>
<text x="40" y="264.5" font-size="12.5" fill="#59636e">Rewrite</text>
<rect x="120" y="250" width="560" height="20" rx="3" fill="#f6f8fa"/>
<rect x="120.00" y="250" width="84.18" height="20" fill="#6e7781"><title>explore: 23</title></rect>
<rect x="204.18" y="250" width="10.98" height="20" fill="#9a6700"><title>edit: 3</title></rect>
<rect x="215.16" y="250" width="117.12" height="20" fill="#1a7f37"><title>run: 32</title></rect>
<text x="342.29" y="264.5" font-size="12.5" fill="#1f2328">27 steps 路 58 commands 路 21 min</text>
<text x="24" y="306" font-size="13.5"><tspan font-weight="600" fill="#1f2328">openfoam_scalar_transport_boundary_probe__02</tspan><tspan fill="#59636e" font-size="12.5"> 路 Physical science &amp; simulation (OpenFOAM)</tspan></text>
<text x="40" y="332.5" font-size="12.5" fill="#59636e">Expert</text>
<rect x="120" y="318" width="560" height="20" rx="3" fill="#f6f8fa"/>
<rect x="120.00" y="318" width="58.56" height="20" fill="#6e7781"><title>explore: 16</title></rect>
<rect x="178.56" y="318" width="7.32" height="20" fill="#9a6700"><title>edit: 2</title></rect>
<rect x="185.88" y="318" width="175.69" height="20" fill="#1a7f37"><title>run: 48</title></rect>
<rect x="361.57" y="318" width="21.96" height="20" fill="#afb8c1"><title>wait: 6</title></rect>
<text x="393.53" y="332.5" font-size="12.5" fill="#1f2328">56 steps 路 72 commands 路 1.9 h</text>
<text x="40" y="360.5" font-size="12.5" fill="#59636e">Rewrite</text>
<rect x="120" y="346" width="560" height="20" rx="3" fill="#f6f8fa"/>
<rect x="120.00" y="346" width="87.84" height="20" fill="#6e7781"><title>explore: 24</title></rect>
<rect x="207.84" y="346" width="7.32" height="20" fill="#9a6700"><title>edit: 2</title></rect>
<rect x="215.16" y="346" width="91.50" height="20" fill="#1a7f37"><title>run: 25</title></rect>
<text x="316.67" y="360.5" font-size="12.5" fill="#1f2328">23 steps 路 51 commands 路 28 min</text>
<rect x="24" y="392" width="11" height="11" rx="2" fill="#6e7781"/><text x="40" y="402" font-size="12" fill="#59636e">explore</text>
<rect x="120" y="392" width="11" height="11" rx="2" fill="#9a6700"/><text x="136" y="402" font-size="12" fill="#59636e">edit</text>
<rect x="216" y="392" width="11" height="11" rx="2" fill="#1a7f37"/><text x="232" y="402" font-size="12" fill="#59636e">run</text>
<rect x="312" y="392" width="11" height="11" rx="2" fill="#8250df"/><text x="328" y="402" font-size="12" fill="#59636e">install</text>
<rect x="408" y="392" width="11" height="11" rx="2" fill="#afb8c1"/><text x="424" y="402" font-size="12" fill="#59636e">wait</text>
<rect x="504" y="392" width="11" height="11" rx="2" fill="#57606a"/><text x="520" y="402" font-size="12" fill="#59636e">other</text>
</svg>