From f43de4c60c5fc76510ae209eb78490df046f2ce1 Mon Sep 17 00:00:00 2001
From: Daniel-rmc <2459944653@qq.com>
Date: Wed, 28 Oct 2026 16:24:32 +0800
Subject: [PATCH] docs: refresh bilingual homepage for current main
capabilities
---
README.md | 159 +++++++++---------
README.zh-CN.md | 151 +++++++++--------
docs/images/current-workspace-manual.png | Bin 0 -> 328978 bytes
docs/images/current-workspace-model.png | Bin 0 -> 372993 bytes
site/README.md | 25 ++-
site/app.js | 146 +++++++++++-----
.../workspace/current-workspace-manual.png | Bin 0 -> 328978 bytes
.../workspace/current-workspace-model.png | Bin 0 -> 372993 bytes
site/index.html | 58 ++++---
site/styles.css | 67 ++++++++
10 files changed, 391 insertions(+), 215 deletions(-)
create mode 100644 docs/images/current-workspace-manual.png
create mode 100644 docs/images/current-workspace-model.png
create mode 100644 site/assets/workspace/current-workspace-manual.png
create mode 100644 site/assets/workspace/current-workspace-model.png
diff --git a/README.md b/README.md
index b54ffd2..368bc46 100644
--- a/README.md
+++ b/README.md
@@ -2,13 +2,13 @@
**[English](README.md)** · [简体中文](README.zh-CN.md)
-**Run robot manipulation experiments locally, and inspect the full visual control loop.** ManiLoop connects GPT or learned policies to MuJoCo tasks through a browser workspace and CLI: choose a task, connect a model, execute actions, and review what actually happened.
+**Closed-loop robot policy evaluation across simulation benchmarks.** Designed to support fair, reproducible comparisons across policies and task environments, ManiLoop brings built-in MuJoCo scenes, LIBERO, robosuite, and RoboCasa into one local browser workspace and CLI. Start with manual control, then use GPT, a compatible vision API, a supported local LeRobot checkpoint, or Jev text primitives.
**[Get started ↓](#step-1--open-the-local-workspace)** · [Connect a model](#step-3--connect-your-model) · [Observed results](docs/RESULTS.md#english) · [Project website](https://daniel-rmc.github.io/ManiLoop/?lang=en)
-
+
-*The actual local application, shown in its current Chinese interface. This is the working simulator UI, not the website’s recorded replay.*
+*The current local application in its Chinese interface, previewing the built-in Panda pick-and-place scene. These controls operate the simulator; the project website separately offers recorded episode playback.*
[Watch GPT place a bowl, turn a stove knob, and move cream cheese into a basket](https://daniel-rmc.github.io/ManiLoop/?lang=en#skills): play verified episodes, pause at each action, and read the model's recorded action notes alongside the controller feedback. Full videos and execution conditions are included.
@@ -16,23 +16,28 @@ These three recorded GPT successes took **5.5–11.1 minutes of episode wall tim
## What you can run
-The workspace brings **camera observations, model actions, controller feedback, and independent evaluation** together. Use manual controls before connecting a model, single-step GPT decisions, pause and resume, compare before/after observations, run CLI experiment matrices, and export complete LIBERO episodes.
+The workspace brings **camera observations, actions, controller feedback, and independent evaluation** together. Use manual controls, single-step model decisions, pause and resume, compare before/after observations, and run CLI experiment matrices. Complete control-step video capture is available for LIBERO in controlled timing; other backends provide decision logs and review frames.
-| Environment / task | Available operations |
-| --- | --- |
-| Built-in `pick_place` | Lift the red cube, place it inside the target region, release, and settle |
-| Built-in `push` | Push the cube along the table into the target region without lifting it |
-| `libero_spatial` · 10 tasks | Select and move a bowl based on its spatial relationship to other objects |
-| `libero_object` · 10 tasks | Pick different objects and place them in a basket |
-| `libero_goal` · 10 tasks | Goal variants including drawer opening, object placement, plate pushing, and stove control |
-| `libero_90` · 90 tasks | Kitchen, living-room, and study tasks, including drawers, microwaves, and stacking |
-| `libero_10` · 10 tasks | Longer tasks combining multiple objects or manipulation operations |
+| Backend | Robot and available tasks | Integration scope |
+| --- | --- | --- |
+| Built-in MuJoCo | ARX X5 / Franka Panda; `pick_place` and `push`; `tabletop_a` / `tabletop_b` | Contact-based tabletop manipulation, manual and model control, independent task evaluation |
+| [LIBERO](docs/LIBERO.md) | Official Panda; five task suites | Official initial states, OSC controller, and success rules; selected real model episodes are published below |
+| [robosuite](docs/ROBOSUITE.md) | Panda; `Lift`, `Stack`, `PickPlaceCan`, `Door`, `NutAssemblySquare` | Five integrated tasks with camera, stepping, reset, and evaluation checks; world-frame OSC adaptation |
+| [RoboCasa](docs/ROBOCASA.md) | PandaOmron; `OpenDrawer`, `CloseDrawer`, `OpenCabinet`, `CoffeeSetupMug` | Four kitchen tasks; arm and gripper control, zero base velocity and torso delta; no navigation |
-The built-in tasks support **ARX X5 and Franka Panda**, each with `tabletop_a` and `tabletop_b` layouts. LIBERO uses its official Panda, initial states, OSC controller, and success rules. These are available task catalogs; they do not mean every task has been completed by every model. Changing the instruction text does not change a task’s evaluator.
+LIBERO's catalogs contain **10 spatial, 10 object, 10 goal, 90 `libero_90`, and 10 `libero_10` tasks**, covering object placement, drawers, stove control, stacking, and longer operations. These are catalog sizes, not counts of tasks completed by a model. RoboCasa's four-task interface checks use layout 11 / style 14 / seed 0 on macOS / Apple Silicon; other scene combinations and arm reachability need separate checks. Changing the instruction text does not change an upstream evaluator.
-Policies receive images and robot sensor state. **Object ground truth, simulator contacts, rewards, and success signals are excluded from policy input.** Independent evaluation can inspect simulator state and stop an episode; a model’s `done` declaration is not the success criterion.
+Vision policies receive images and robot sensor state; **Jev receives explicit text and proprioception only**. Object ground truth, simulator contacts, rewards, and success signals are excluded from every policy input. Independent evaluation can inspect simulator state and stop an episode; `done`, controller `reached`, and completion of a command sequence are distinct from task success.
-[See real manual debugging: arm and gripper controls, sensor readings, execution feedback, and independent evaluation](docs/images/workspace-feedback.png). This screenshot shows feedback from a manual jog, not a successful task.
+The local workspace has three panels:
+
+| Panel | What you do |
+| --- | --- |
+| Manual control | Translate and rotate the TCP in fixed or tool axes, open or close the gripper, and submit a combined six-dimensional increment. Inspect position and orientation errors. |
+| Model execution | Enter instructions, choose a connection, run diagnostics, and use single-step, continuous execution, or pause/resume. |
+| Scene settings | Choose a backend, task, robot, and scene parameters; apply the selection to rebuild the environment. |
+
+Stop remains visible across panels. Switching panels or between picture-in-picture and side-by-side cameras does not start inference or advance physics. Manual control requires the current task, action, and request to have finished. The manual target-tracking budget is separate from the model budget; TCP rotations are not individual-joint commands. See [workspace usage](docs/WORKSPACE.md) and [six-dimensional manual control](docs/MANUAL_CONTROL.md).
## Step 1 · Open the local workspace
@@ -59,35 +64,65 @@ py -3.12 -m venv .venv
.\.venv\Scripts\python.exe -m maniloop demo
```
-Open [http://127.0.0.1:8765](http://127.0.0.1:8765). Select a robot, layout, and task; apply the configuration, try manual arm and gripper controls, then reset. **No API key is needed for this step.** The server listens on localhost; keep its terminal open and press `Ctrl+C` to stop it. Use `--port 8766` if needed.
+Open [http://127.0.0.1:8765](http://127.0.0.1:8765). The workspace starts in Manual control. Use Scene settings to select a robot, layout, and task, apply the configuration, then return to manual arm and gripper controls. **No API key is needed for this step.** The server listens on localhost; keep its terminal open and press `Ctrl+C` to stop it. Use `--port 8766` if needed.
-Continue from the repository root after stopping the server. **Windows:** replace subsequent `python` commands with `.\.venv\Scripts\python.exe`; join multiline Bash examples into one line, removing trailing `\`. After installing `uv`, add `--uv .\.venv\Scripts\uv.exe` when running either setup script. No activation script or execution-policy change is required.
+Continue from the repository root after stopping the server. **Windows:** replace subsequent `python` commands with `.\.venv\Scripts\python.exe`; join multiline Bash examples into one line, removing trailing `\`. After installing `uv`, add `--uv .\.venv\Scripts\uv.exe` when running a setup script. No activation script or execution-policy change is required.
For headless Ubuntu/Debian, install `libosmesa6` and set `MUJOCO_GL=osmesa` before starting Python. EGL is another option with compatible drivers. These Linux settings do not apply to macOS.
-## Step 2 · Add official LIBERO tasks
+## Step 2 · Choose a simulation backend
+
+The built-in scenes are ready after Step 1. Choose the optional backend you need from the sections below and run its commands from the repository root. Stop a running demo before following another launch example. Workers use separate interpreters; continue launching the workspace with the main environment's `python`.
```bash
python -m pip install uv
+```
+
+### LIBERO
+
+```bash
python scripts/setup_libero.py
python -m maniloop list --backend libero --libero-suite libero_goal
python -m maniloop smoke --backend libero
python -m maniloop demo --backend libero --port 8767
```
-Open [http://127.0.0.1:8767](http://127.0.0.1:8767). Select a suite, task ID, and initialization ID in the environment panel. IDs start at zero; `list` prints the available task names. The default is `libero_spatial` task 0, init 0: move the black bowl between the plate and ramekin onto the plate.
+Open [http://127.0.0.1:8767](http://127.0.0.1:8767). Select a suite, task ID, and initialization ID in Scene settings. IDs start at zero; `list` prints the available task names. The default is `libero_spatial` task 0, init 0: move the black bowl between the plate and ramekin onto the plate. Installation needs network access and several GB of disk space, but no demonstration dataset or policy weights. See [LIBERO setup](docs/LIBERO.md).
-The installer needs network access and several GB of disk space, but no demonstration dataset or policy weights. It downloads the pinned official source, assets, and a managed interpreter while keeping dependencies separate:
+### robosuite
+
+```bash
+python scripts/setup_robosuite.py
+python -m maniloop list --backend robosuite
+python -m maniloop demo --backend robosuite --task Lift --port 8870
+```
+
+Open [http://127.0.0.1:8870](http://127.0.0.1:8870). The five Panda tasks use robosuite 1.5.2 / MuJoCo 3.3.7 in their own runtime. This adapter explicitly configures world-frame OSC and records that choice. Camera, manual control, and model interfaces are integrated; no model success rate is established for these tasks. See [control semantics and checks](docs/ROBOSUITE.md).
+
+### RoboCasa
+
+```bash
+python scripts/setup_robocasa.py --download-assets
+python -m maniloop list --backend robocasa
+python -m maniloop demo --backend robocasa --task OpenDrawer \
+ --robocasa-layout 11 --robocasa-style 14 --port 8872
+```
+
+Open [http://127.0.0.1:8872](http://127.0.0.1:8872). The installer pins RoboCasa and its compatible robosuite source separately from the standalone robosuite backend. Its five required asset groups occupy about 11.1 GB unpacked, plus source/runtime space. PandaOmron's arm and gripper are controlled while the base receives zero velocity and the torso receives zero delta; this is not a navigation interface. Default-scene interface verification does not establish autonomous task success or arm-only solvability in every scene. See [RoboCasa setup and limitations](docs/ROBOCASA.md).
+
+### Keep runtimes separate
| Environment | Purpose |
| --- | --- |
| `.venv` · Python 3.12 | Main application, UI, model connections, bundled MuJoCo scenes |
| `.venv-libero` · Python 3.10 | LIBERO simulation and its pinned dependencies |
+| `.venv-robosuite` · Python 3.11 | Standalone robosuite simulation |
+| `.venv-robocasa` · Python 3.11 | RoboCasa simulation and its compatible upstream dependencies |
| `.venv-vla` · Python 3.12 | Optional LeRobot inference and learned policies |
-Keep these environments separate. See [LIBERO setup and task selection](docs/LIBERO.md).
+Do not combine these dependencies into the main environment. The RoboCasa worker is a simulation installation, not the upstream training environment. Optional assets, environments, and model weights are downloaded locally and are not bundled in this repository.
## Step 3 · Connect your model
@@ -98,12 +133,13 @@ Choose one route below. Stop the existing server before running a different laun
| GPT through Codex | Official Codex CLI login; select the Codex source | Uses models available to that CLI/account; ManiLoop does not extract login tokens |
| Hosted or self-hosted vision model | Enter the service base URL, model ID, and API key, or import provider TOML | Robot control requires **Responses API + image input + JSON Schema output** |
| Local LeRobot checkpoints | Select a supported model loaded by the isolated inference worker | Three reviewed LIBERO checkpoint presets; not arbitrary model-folder import |
+| Jev text primitives | Enter a separate TypeSafe key in the model panel | Explicit text and robot proprioception; bounded arm/gripper commands, without images or object localization |
See model controls in the actual local application
-
+
-*Current local UI; model and task settings here configure real execution.*
+*The current Model execution panel in its Chinese interface, with the built-in Panda scene loaded and no model request started. Its connections and instructions configure real execution; this is separate from the website's saved-data playground.*
查看实际本地程序中的模型配置
-
+
-*当前本地界面；这里的模型和任务设置用于真实执行。*
+*当前中文“模型运行”面板，载入的是内置 Panda 场景，尚未发起模型请求。这里的连接和指令用于真实执行，与网站使用已保存数据的体验工作台分开。*
4aSy8kWk4|x;{K-m)!LUu#YG1t
zeq==C$CmJ;&{WtT9!F!&D@S0?-#dtY=TfLrT3#lQkAt=bFfxib=Sy~;ek!DV!l%3L
z(P5ts!jlSLc@T2D0R;z@V@my5lgI9RDonQhjazFg8ZoUvSXOoFeVDu7wlZ0~%H&nRgHWJ_`2^$LRH{G7dZVTI#*Y4-U3VRHVeR1j(!ay)u~@r|
zw@=bfUfmGxlccJx52$ffWi4w!7T|tNe7nC}1*k>k_wlSz68VuU2)DiY+Kq$Xe`9ZP
zMs}tu>Mi=*klZJrf}r>H(dT%Uv-fc(ln*CH8H$~xrJBJc`=u#tV;u-=Wz+6BS4{v^
z0!MR5_{uU$v0E2OYN@zH>4qUA#3QZ#aj+pNPLH<^JZ3L7TrWcBE*oGM(bkk2XuMhh
zTcvs{casW@VreB=W-oCGJCOHKCe?~1UfY5h3QCF(jGap2-0}`Gl=Y<+Bu@ZZyrsBq
zljGa{O7dcg>4XNpBjWwPv-n0qE3%ZBnD(_WKOVbWTgj#jqtz8$@Kw)>@)E!N^FWg9
z{z@%Iv}T6;NHK0W##l3-$@!pi3rNb*Pof|(c%T9Hg4D05iY!W~(l`3h%=H)Xd|I&y
zTXYrlEi!QrY8wYULwrkdzp6X-<5E=+JBW$8_@`zz6m_hW4pqvQ(AjqyU+5NG1tKAErO)Ax@e+
znkT?MqVYpj6g8x^lMgYgU8DW0(M`l0ox4?!GSs~dSPwz68kfLOcb737aVaSjH**Ow=t
zq+}{Fy02fNch!ue6PS!&7)8UB+6x&{Zk^ot<(=<0EEV?5Lq4G;G*2XCvDlmtrJ6ylRm5=v@{W>)
z--Pa8ri1X8GVJK6fhIm`zXx+y=mdO%WkfK4BR*g*J=$Ij@*fL}wg9VdM@Z{FS?
z<{kfVyqxghKQ=LAeGEl%Bv{%knGHdWN-8@A;L$5rSr7ch0j>?M@pM}@KRp~=cIb3B
z;CpR!?Qh2oAgm@5*1Vmm1EXqZScc;fa>LHERpw)>m7PGgv$lF`u;M#)JT$?$_T-n-
z7N_~_Pf{eYwy`v;G*^XxZ0H}|_OzSH7lL{!C087mGWtpHDi3J1PLEF0_dJo8A;Q9q
zfFADIdMx^%@Edr%iS-Ey5)_Gq){%~qgm{PW=0ZojEym&9@BQ;y2$TLa<)+!t|7HQ)
zB16((^R@9yoxo>B?HAZ7-A^eUPJoywf_S;B(~{m`Fc1(I^{+f@f3|<*jLq=}{Mz4%
z27_F18)88ex9+C{zt6ypU1^i`q^9)JvwdX~tz!X(k`FK|y>V9n6VPOykG>HN&@lat7haNf*LiP?WER~U$
z&7f#RA4b8d$K;}kVTBZ$b^zW9cs5ExD6~SU*i`2%Y;Sk&r}iKJuLx9*QjOW}x@X1Q
zAp6~Hw#k9-O7V`k=q+rqDQ3#IfC@bnB(Wz^GPfNbq-A3#g90-Wh8O{&`A6C
zWOvNrvSjE9Zf)i3LN`b*h*ljk9QN!Qb1pH|1M;@@=&B*vn3v=(tJkQfJhGPgs&oE8
z_nNJOusovUV9Q?c3)|CE;NUiFm7Y$-AyJx_M44855|$`^MTTrcl!02X-fNrgtPY6X
zv!6#$tljgntuCgKiCSwJxuN7lhw&wpavB@=DEL^uO`Bpw7@Mq4%@O}|d))KH=S9JV
z*;mIhbM}Qz(JYT!WAQwCyNqSl2=rXiH;1!s*5Y|&j|&xj|uX+MsGW3S!Zw47{F)5
zHe+o3eEBhf+hv_!`8+tc(CpSvquvG7|Dza)U=a&C0$4*DS7WtT4cS??
zAH6AyU)rdehYYK_M{
z(NS|-vo;Mpl%Sokhsa(;!lYIZ-gni;(31P@ekiRX|Gu1gy>uMSK4~8a@rZc;b4!2n
z6eU-efBUiY-+yxNp2)#}zTbOaB%lAE(7ON5Jb(;W6#jzp-+zc9`b&QN_v8P&_37Td
z|EDkU@Lmkz|Lwbd+kdU8$YMXtVjroC0i4!9=CSLcbH^b?X>DyqMIpt&&Sh%qUpa5u
zfkF(lNeA=~7$j)TsZ?5Wvvc2$taARd|Gtd^R4=&6YJT`j;)sNZe|Tikq4Ckdyh;(q
zD>$Q7&xc1Mm7=g6931```kt3&X0`MNv+<|p7UiU*#3Y1BC4tW_&y73t1P`v}=om?W
z9}=kc_bWC_%1QP8?$5~*q05E8`9}aGpy=)LIyyR1b!<>Z_M52NKd)!z#(8)s)~0$y
zD8b)%!58xU=6g~SD)dyeWSh8uKM{_Y_;<&muG45V@CSLY6%wM!P49iW9q`W&c26|J
zNEB7VoZIFX4D~*sSc(P@#DO?q;7ECiZRF{t0rZI)8nlF)Ty=HQet>)>r7OYCPWvwc
zhep@Uf`ycmqWJ-UzN2f<(yD0G70v!!Dcb1`wpu%*rtOHsi5gUh{4OIS+t;N}C+0Wn
zCod@`D=EeSp-T8?HbtWgP6i!s0K5V?ZbQ%40B&84U3rI_E_B%pe#)NiqP(tuJcYM!9A`g@hxL|H
z(L%<6lMaNYtf>u+j&s>>MWzU|ujRh4*_$MDgl{WW5p9+?%z9K{dMl4qHz&jl-!$)W@T|^oAEK!hbs^;ITFz1U&}nDf*TJj
zDct;hN$zNC=FhT*Z6Kd2We4fjH6r+Q;=e>;i%
z=Qq9jEtGh3BjVyoGLNI(?xc#K0y7JRHlW(*3fi?z$Kc?~q;Q4#lW^6>btK$P1j^Zv
z_os0QuJ-X@0C%RM0Z46pEvPs7JzhgyCV^Q^LxYNs?*?q;fOX9@01&|({|sW<+aGFZ
zB#5K${BXY_N*f^Oa#jQy*kg@bPyRwR29Q%m=ZcPt6H_oz)L_;bi`9`q7+&XDPodzA|`o({|Ssm!}nWMSexnYp&_ZO5aRb_)W`t5*>&-?!1MRkdIav(7ZUWH(Waw%
zhkIl9G{dcr+k%K+I$Y|o5=)z<`(8ILAWClDn5Mot)P9sUl?y9ZPnID$!Aa%g~lDY`;Ltb^I^d-YTl@8m`j+oI-gCa2b9wd+!!h>2JvQsV*0Y{x&ff&g?wFX+
z;^rUr>L40jU-PQ$P`94+=YKqrzki#NQ+AsueyPz0D0K-*$BXlPv|j0yyT9`o1R71|
z(-yHz_OiM@p-V1=Lg2a~I@ouxQn493$QoSge9W&sT0jCve7N9}_rbV3>K4=1)tj)v
zyQ{~}uA=wNaeiezpF_oJT5Y>
z)~E}g7xt@1p!u2}31&q_mqxWY&IP&dX!x+=77c?^ksR%gL`HQeTPzt_r$(){hP{uG
zp{bzVTDL>Kbo%GchrXT~kTcIJH4Lc9Yz%rmeLxUtX2<94s?ucb^&if*O)`yhowa5r
zx%czYSJgf<(f1pMCWzJS3DnV=mgd#9iMkd
zT1DXs{TRn%1q4DWvt^ow;vKyrzeH~D-I;$l4xMg>Pzr_r`+6K5F9FfBy2cTLt9yHY
zJE4Yht!V$4PI!wW
z0l~t=q%O`@S?FVptQsL{2g#@3a$7tJ`21Q}~jZtK;WMLhP
zdkgb~BBX4c)%;aiIgXyT=y%-7)y>W4I}1}&x*fG3QL%31DzrO`W7U(^??892u4AKP
zs%xrfzXHIysoahY9{29b$`lGn@OQ#&XrR>=`a$;$gj9pi#qP1-2TTN#X?z|_I~|xk
z%-xfd9Jn{h%<{T68JA!7byQ?z
z{lf%l7h1^~fhI>$@a8)_3;Vn@_NBju^&5-MWCq8%E^t13M-vE&7UT9%H?PUH9mcJQNlv$vZ
zt*k|cF}cTX0YQAw?2JjSG5z|zKT)AKVPbcwg>v|79g!FKLjoCm2ti37o`GCzt@SVx
z%}~<4N`Lx4l?);8&GFAwQ(0-RR5TqGt)lDqxQMi;L`{{H
zpvGn~zqM#urb$Ov0%pl#B&ncsK&k>5DOlKtRQyl1rTU|1B}-x
zhFf^}W0?$u&dyJ*7qV)S0*pubVC%mdC4o4vhbCEAWS9)osHlnlAiJS-iLqa@a^$gM
zAeP+Qp+ToeN6sgTRvu^nw97*(xo)A_iqT)-q7S1WzU#R)OG;UJtJEr4I3m5-Y5Rq{
z`+bj6twR1RWBPKlb3>Q|rTf?<$m(i#z1jh18jvx8s4oUoF)69}RwJ*|svgI~`QFXj
z$tVlKpoysoodPU+YpVqL)Y`W
z+`oLZ9E>i*cUFq7*IP=Ol58AH(AoR$D#}UY(R^EziTbrV8!e0FT3&9I^hYnCIs#!e
zQaETwiv=Y`9Qs4A+{e_PJrY-ygDx7*pR+`smfzWEw%lBtdgbxFF4A<`oy$eQ0_v?L
zsmgjhGB?sN+Iil;Q$b|s*!0kPCp0UI(g{sO1yBqR1pLEq#BlH$6eaY@!UKe1US7d{
z?{mx}P8>RnH}Zu8=DXEyMHDHk)Xo+GHPuC+Q!*bqc_9`A-POzaN3t4;+B-q;NYSAR()Rf*SMHzq17L%oktCp
z$=5rYLi7UuCpR2&dIqow@qIPIvttH_+Mm!UZ}r?i-Ic&3hs<6aA0Hz+IAW}mVJ|E#
ztt^cm$y+8%dyt+%PK!ln8Kl$tZ?%EflOfzh(Sj48eof^Bmv|{E=9q%0DUTm$h%{Lo
zr$9)u%J>2fo`T}fq%ZY2CC6-#L14GA2(yuZo-B1Qoqf+YcB>V4V4d_2ygr_fr2iOm
z_lhhp0fm8#o}P*+qrE*Q=c7B*X4ofLQ_;5yjd--1i-T-%AP%Z5D6+M&_#&C9gP+_P
zzcA{rzD?KJvl^nLsO0cnDX}m)bw9(5v5MErEcwZAfg$H^%T`0j2(J(*UVEoBsU^1|
zh1f>W+{k?Kf+aK{?^Yn*n<|(DxdblO
zP!trDgXyNSozFnY@9FJ5Z3k*t5Du$V?`n6sSF`jwvGaZw7Z=wvxj0-p`BRAyWmHtu
z*>&NxrCa~sw!awf03E6fc#=!hmBzb-lTZP8O#wMnuINohkc+{1#V$spF
z`xE89=VW1d824>zXxMbK38RcSZt`-ZKg(^Zvw4Mpz|w{|>tX{K^?2Yyg-dP6Y~?mT
zSB>H!#qKPsEX|-2E-hsMODD$MikqGur*t7DbmN&;T}~#7FQuj)OwqT_%w=EM;N}Ky
zw5=7lv-Xx#N_2{`P(LAg&D;;v)Li~4%@A$cqW6lWFo2q&43@E@C{grv7Xb_oU;$&g
zxp7jj;mb%%=Oy$5$n#&C@rBMZody703~DV>8BM#;9dz`{Gq1rP899uj*Xj)WLY)v9
z8M&ij`a{!4Xdc_^Eu3m9mvaJ>Uh|GAkWl{|1rYtNp&>RtJ`xIwyn1vLR8+tc?Ck86
zNcid7@)#IFJkRp@7GOSX*C!(g9{;$xyH7Ai0ZDnTzkgF8LfG@wGB-R-2JXcq2|0Oe
zTpYwrSwJo_O<2TBMNU-EwXQZM#!%SRK$%>qyS~2uCE3_04u?QepA9;5v&VT9k>n-@
zAy$LMqHvxj;9O(Q_Srojfd#YPbZ^vBD*>RaM
z3)QA4rdbpfKR*@hc+AdBj91IyECtuKezz8tuS^O>iFLx<*y%8Fwe(Sv(&&#G?)QTb#H3@gQE3S24p
z?B*Mxc$2`CklR*z46P)%2a%$*&)PNmn^5*bmDyRGB}oNXe~X}XVZ(Fig<>!*{LZ*ox_P50vtI23XNUiHaTjhRVkgAEYaYw
z_+2zxko>F$uy+7Xxj#|U1UPHa6k#flE%trC^bpXF>Rf&gz_df60E#KKg-iq7{jDd8
z8+M8qM$>t{&iaOy^lyGwROLxk)>bGfC={tRg_ov1<(`|Ey<%iU6R)KDWj^*>tmxC>
z)^L(AY)y4FrA+efRR=pC-_u#Q!$D4uRLbAuZI|62t6DTPg~-^9={(-RR%r+VZAVgg
z{`F|u`7;=34GObjl9S_OqiGpQBBPV9Il2p@I$5A-_&?LWkn=0@nHq1}5a;@@}bVwhRnW
zKo9xb3*|%iuu>hq{dJ}C0)5{q;rs2*DyZH}~hywq2Kgzuyw4GZ`0WAOUJVh=|kxp#XH0lp_9s
z;XK|xRbBlZ_O7U)&>pMV$+6nwYDa)1zPb4!CkP`v{WBf`L6g-YHc~g(pZ555b-xeB
z!kn1&4G&9+zf+Wa8~--H`tESi<P$?U$8
zzbF7?(fWG7gallyxg<@0K-4j~InfI(Kf}Q?9ST4t80*6{Sgv=gf;@X)L?tILKHj6@
z!rb`g`8B^CTz0zzQexPj*N_>au
z{NZs7ELX?%ynlPdT!R5Os#>k(ZdYIgCgw&1hq;-#v5_$|SY~pp6V_a{Rkg7K^(teJ
zRyg`ELrlHizkd&8{M{PP2g~)=AaacqHPgZpHW1|*x)&=KiGDFP)>rSecm+Z=>D6n-
zr>7;P#3HeY-cCkJ>I?qJ%sgG@=0H1umSkrSN;9#2G;Ghv1owSd1ho5yd1y~hTzFa1
zCyUeDs`S)UUn#6be+QGof&xa>X3yi&!*vSj`3fhDf1ToINm+e)anNbW5+wVog<+KY
zvaFVg+4DW-9Rh4E%{KJJkp)wez4J>QM-wCESs@9IOHbSyof6djj3t~
z!cD&>C*kAaWar_0wpXE~og6xNpY%Zi@gxx{AS$~$kAdLZ_E4YP58^_$f-V@gbZulH
z1Z=-$0@V%-dc%K|Ji7VYGJ?k~mDfRBO-&3a@R2Ni1?*0LTwWY4H$j_Zpnyi~=l6Al#o_Ugo||
zK72mPPSyyC`=RmWJ(uh1df@eiZk0V!!-C7sw>I@iIXQqjWIQ9o=P=t^RUMwb^o@u>
z!`cy8XVk8YK?1=3`-}dTGzBl=v5j(W^2>b=&wU<$KKF$SJn+1#&0Pc|UBnM`hvgOJ
zk8YnkU0xXa^()3(^X{(Z2o=plbrK1J@Rs>-0YCyHB_q|;g0^_vQvgXprPCb&I>~Eb
zmO9GIED(x6HJ^22VXXu47m{LPrTQk%9hX|7qD)dHUwawA^d89U!h?d;F5ALD(=9eS
zI+@9grX@a#u92z#Kl}L!GfO~;G1!&!K|Y!$`CZ1;_fgtu$0R(0sc{U&E&GsQ&h@o>
zG1v1BczCX0d8oTNhlh?z9sz6d+Q*|^(sd}nI;sM(|j1e
zV;M!K**3H7nwSUlEjsy{VOC`X^C19|iGhK!KauP3i;JIOVMZdV^}2ki2e+sE}&c
z(=333ELD3f>G+UpXgwb){uTy;|EYyW6Pui#p4aHka-dK+9M8tc>T87iC4u}E2uxtt
z#$^KDd}ym1JvH^|)AAx`11Bo}Cl-!6UE-Kp2aXa6(Vr1PLDE1wzC#0&%~z`+!oqjw
zCNjUjC^R`=+8esMN;*z$hHQT-)mT3~6l*iimm$>oY_s*OT$W)#`EM{X@Kx15*bg_I
zlyY1|+1lFkwB)r=%bSq;jYAp#-3Pe8{hSQtpc&!4Xca&SA$(lU;Y
z8ZQZULgmE8%>kl2^yKQt)z#HS)%x1n+Rld^+4-^d_KATgBFp7hUS5ytHRi^KhOVE`
zO;rEa3vhD-$KGTal9a6D9lbgm1bAAAE-Z`<47%cWG3{2CI2G$%*E|7{6@cx(8YJNo
z$fSPq&2m0!@r6fIt2ncPhk2q~(-shl6Uqe{I|(dlPfzTSH*hl1`NIp1W`vK}hbtA+
zf2NAXs<7dDw$E?gMMWL>USEo@bRk@Wkc9h7VoYr8z4gP~q9T!s_?;%ImsQo3d1hRn
zm6Mxv?SN@|$F;ywxT3s#plB)H7EA%W$A{7VhwJeuK0_{-R9}&
zI+_>I42+ou;U!uoj&AQ#_aS`zsC9A%!NL|$sG3x?urg0bNsNt5?NYRk{f$=qcWhID
zkk=P{gU{u##hqX?-?2T+PA0gI8M%{JP{3Og%l2()NlL;eYu?V-(TItS-MuN%TC4@t
z1wmV!kBCxA;zDyBO!(3v8$08C
zEglLtmvXM#_w~@V(g~Oo4B@c%9k01pt7>cYJ-P?_`}F`mi%#5C&)w^^a_YYjBr_p&J@OmVnYIS%&o^K^BI9PnK4kys_=;-}hbg3XESwVX(-X2Mz{<7cU)gS-dIC6@MOVU+MywZpT
zuq0m^A@a?iCu1+Wb~R6Svzx%>_zqsULxBGXK4h}BrK+?vyzSY<`t_!Gf2X?Q>bBi+Mqj*h~y$f7^6Kj9!otQIKi9w(%f!M4kg=$QEhRc_<9(uI4%qSjFWS=$J08IO&>d4+Dv`yk
z=VI5{xfTx(?;;yOhsQx)2_c{7&BHBnAR6In5Ai2YKU2!1z&FI1s(^CtujpK;65*}7
zLm6^d4#@Z>0P+gJ)$>g*(1%+Gv%Zvhhk2@$m&JgQXkR0R@CN_kVSbWE$ui&#VW;wg
zVHh4S4G9VQ3+#4c3JRu&p<+Vvh-?rs)uv}4h8VO7c|Hcs@QYRLv&+*oevd|gzr7FF
zQdF95ygcx_X>WCxzHD;bPyCet2%0B59lVVnC_f)fa9muS<wSUsnygF
zcN5v9$`a~3iDy@qzKN;74Zy^+d8DP0^SL~`&8{VEWH;3;SNG5teDL+)V40ZI;qm&d
zvnQ_VNd$S~>7a%tv0Idsl-zh$B!S!@A1rKaVOS8Ku(@}qH#e4f@s2mOO*b<>KE9%Y
zdC?0c^~=QT2qONh>N
z5FdvTaEz(xY%^#bS{_opdWJ5rPv(vO3=YQj8VueXjEYwHol(kA&ukU?_I{$;Ou^Lf
zZkU-3)6sk~&uJf!ja8#5r4FX6wm~5sT-+!2y~)=X7VBskQf_fgbf=Xym+Dnu-Es#C
zAIgG4vz=W6%lYb+!)yy{xl}b#F^S&Z-r;0+dQ#G*3UenTtZAwJ*@T27wZ*wK9&f|-
z87=NwYxm0mf64=btN!Rfdi6TD%QGSUtkoMO4Uxcom{zae5rZD7i}Q0(BNosH{!G4(O$1h}4bv^l@oP;x?loDM?hufChsaPpyM)b90X8&Y9$lKE!#X3JblTOEW#Ewo)ZE>H5@(_)7O*&le
zu>+)V=J&KT?e8zXgW$g0DgmzBoGMv<{%3H$S79twnW&^L)ByVaPH4;sKex-#RH2dt
z{?Xdn>BVl<6D+*EW_S4QtPb*SBe~Sfz<~KUsX#Z9_@8=}7poBy=&q1{|!2O^lA?{a66idUtwVQb!^;Ym^$OzeI7lW<@Gvr2CJ)`q*Ff+
zDA(>S)%50-G&Q&IQr{w>+%&r{=K%)GE~&n-a8FQnpy?|JKNK-KA0TAY25Oz(L=f}Z
zx;+8PehJ7y4Hg%7+T$FzH6l0Kg6irUm;5d!r6(t+7J88yniLQR5U9l|55Q7Ew9+Eo
z3+%>_kV~C5SIJrSUg@+t=Xp!#*Sv&0_ADB2P;Y?^KDRWpJB%2Iz~O1CW3Pq)z*T5y
zs0r|rQ+X>OtL3L=1qC+ReA(z(S)F8pzT#I$y+@KS4-+P^wzA{_9*f-GgbklJN
zguCoj)#!w4)8`_9hBh_MVLn?S5~=XH+QmURVAcMv&yMx#cp;;-^dBC{`0{q$1IT81
zUHJm4K^#8joA&bifW+dD$WnyL1&Ni%w2X8f-SsJ16Ml-)RO0K)zYC>n;nZ
zPiO!UeCJz?U}>j9M&vG=lAK(Hxl-@W4z7!f3p|WBhj~)t=9vPh_F@u#eQ86ya&}(f
zFt(b6gi3bK%pWZG&Wf}Fb2IBW?Kk>r)Wm8c&h^@c`9bLv{5
z<$`e)6&2~}>AC)JvyV_lRbiOl4>2*Z7)VMRIXRTsJk~LE3^5ETF@V{VyT{3Cw9n}4
z9WMB?IRTPk?Q6Wyc&C<_sFsL#HErJJbvxC&m`X};g6&y;8v}$0mSO{*wRx
zlhrieN2)I;LGx_LQDg|Z=QpVT^scYIAhx?WCuZc#$;41Qh12bz@p7q(7nu!$F%u)ehf2Kcw#s)
zzueN|*RNlZb5ughS6;zAAtDAvxUFK^Xga6_}ElrTpW7*ub6n1xADJ#
zH&74M72X&Sl>iV8Uy;^j;+yu^ee3#24EFl+?n=j_Bx
zNH7ZIYQx%G++LI^Iyu@xr)CD&FT}aCd9>5NlbKZ_o>lAt%j<|&_b;_AUppq
z_gKVVvqWfx?6cU;-9(FAPnFIlc+@rKeacl-%&;wVx-X?)a5VH|U)`}WcE9_+Odt}T
z?shlFM`pl^KiNAOCVYRfX0^U`sLcoYCJ-M~;9vS(PTd?p@N#0RtHVmZ+RuBu00-xd
z%D<8PzkIiqO^4o{j<{YaE!(3hfpL6Q5Lb@5185XpcOQ0c71aUGN}Yn5`8)cQ*57er
zHBLBI2@Alxm`k8S3Lt{EC8ovYl@yI`pH57ialZ-*LMc0(s+X&(#nhVToYD28$B2Hn
zer^4%Lu~@UShb_=a&8T7V984oW#^V~dmdbPanG0-=iIwNkkF8cSj;OE5>~<4LQl!=
zggQ_g5Sk)MhTzrTR^t3kwXb8Dpn%EoVhTL(?hQ@yZVSy1i!N+dyONNmtM?l~{4J!k
z3{3yrDQ7ITCl9;->(wNAf<5d1Px}xED1+KY5gZ(4eIPrZW
zrJB4rab)U^DP>aLV1-zyH^PIU`Xz;d*y5tk0JL2t18C>Xy!-=w{lL8K=N};ZuyMRE
zS0_+4w?hT!yTJYA2ZxZxZKr#Y)1Xac9VI`(=BNE`{oDb`N$O^PHpD)i=rLgYCV&Ea
z*e_A;1wcx(F;^`$S(WU3fi>VvNE82i
zJZPBz^UY*Cz>DoamKy!97hslnu)+VSJu5fc=-*@>Vs4JV6*se_z-(vtl}rIRrmoD4
zh26iI{%SFBa^aYr+vx+4PY-YRvVvR;Eb5(lAhH@~zP)QYoSBLi`2$T
z@X+9J$clfFmk>FgsW~TXy7GG7g5%Z{Qj%h#M(wZrKG)@+&6p&FhnK4wl`i5@=Y_wlv`~
zNOg*RtMk-y2%eT0vj2QF5;78Ua+3p+GWTS(dVn4e!VP-6zX7jd8~=5L&29%TA%{~(
z$v)1vzPCr<<@wwwe0XPnh7>-%ud2F8INuhAP9c!REbH#BE@@bwm1S=%oL5sq
z){+$S>sL%nduvj1vPj{yqqume&BDYjFg^l}PVfMYxOkz!ckiWd4_>@GBwT@P!&(~3
zy}iOXX}8EG;iBTA_A>Wqctyoo*f6H_Z$VqCE3q=-59lq
znPFP&!m?)jR@T-GaH>dW!Kg{qIV6~Cr>@Q5vYwxp=NAaC)nYr*H_Y|iA(;>4yF72d3AV@l8ry&cv(?N1tc#=#l!$M^x45F
z;rPF{^6x^6;GiHmN1)PzssNGlD-6B32K$p7cp|y@AUb%Ht1|6g^Hm6ZwFm
z+lzLD%V!->U0u8r7@pqviyovdA1qyPfO7_mnw(5aPW=w8)Q_Bzgn^QTC{CTEu6~pH
zw<+_c)aV=y1q)F4-jI!v+nM2Uiiw#qpF;UA{$2cXs#2;kr&?b6hzm(6aC~eJx$SlJ
z_qGIC_AMekpc5o#&T%x=HOC}~Ip*;p6Nf8{ocKqF0=?o+!
z0*GqP1;-5};Isub>-fm^X^~552`PhYCc+{^aH5C=a!L
z{{0;e<^OcWG<5ExZ5gt%OMK?bBxABANQn
zrteXnBppx=1{>y}G#hJ*C*r4Ljv_pze0jOyR7jX^cn
z?U1>4e?op3V11t=1!R`Hvk&!;OKr%pZY%%Zkzp@@`u}AMq3~<#A)etuEjffyCHKVf
z&YGSWl?qe2s(bjulYx%?AZv6a`S#iJ5eI;xK<5&yy^x?aTO9C91vfSfO^iTX5Jf%(
zB@-6fbJjCHWF?bT8Xo3V@B_tz{}eEQ>=fB0V}BVU2g6KnUK7A(i37_OpNr;5Pceyf4-6z+RW1tewlB65Q2Ko
zCn`9gKF3i~aqimpyq+IsGvay~0+b@jJB{d%_eI5pV$awrngRYW$gW0wIqQ#=2JJfR
zHf}~dTW>?dVfA2c2se}-IHzTvKPqpz`9gnHhl1NV
zPKdFI2@3Ku$)Xdr1*{O!bsu{x5aqx|GK}%S{YcAF41VveiLB9WMl+9|iNI&fr|gZ)
z$M+9zui9*|AfGNB_=CWoZS^{*xHR=G&f{UjiHR0)u1ISqR?^!cQp0DweBHz-EhZKY
zwhQbY)xL6I?VzV-mKIaOc!90x!$SG(;HGUuay5QLV;+tD%r*=a#-|%&>zib*V&XRH
zKy-1lyt46Dop;$Q@ZpP7qAHUKNX+
zG>{%i^nv>EgN$vZ0ty%d(Qx9a-G*MoJWz_3twO)t6C65q8W(+b1llIHJDkKPEn}K5
zxkcXrEAK;D1{F2=;DffS&|_eK+so+aNUHaJv?mKL7&qJT=C1lEGCz75_dgBa_uq-M
z2^y7dT{sZ7G8vr@#_^T?L8HGOe$3ZQpd(rO8~aGBwKRqFV2kOhVE5IHysd1<1*eIp
zxoi?#7TKYb{Cfv^EZ4sttE&7bN17xq~u6ddgRtPtgH
zF<_pGSXfy$u#^&zWE)=631K;zna#&+{JE)D&6EF&<=sR3*{>Ia-hr`7NITMic~d}H
zTR>TvAtG`$L=4BXWNvbzS!ecR`H;8yiuex9K4ob@>@R^0EZ&ah^b}2IV$z$RI%8BU
z5uj1pR_fN=To%UB<|LfddxgDGNn
z(%k$vub1{-I=1(x^VHR{qZ>J#s*ue%Pu`!QkZY?bH#o}ZHYqnZ=uzyi>*d$}c{lf|
zqt6#EfhtSLLJ-Ln8HlW!4%He?fgkbh+5IlNxBbWZCtb3hODbrVwjFt~19{{CSp
zCpfI@jItjik?S6+?~Bo4&PcsTbI1GRi{zY}(^FXFSt|iABc!s~22WX=AtGI6rn}%^
zAmQulxXSvt%jT%gmSz@j842;S;O7kePn{9oYlE8Ep(^sNnit0!Y7pdBsj?rBk{9Zw
zyGqW^#_IN-#>VQAI;4oyW?Q3C@yW^2sajVy-`qwzpXa%eVZ?%qB2*S{8kf_W>`UL<
z)@#_ldyP$o829#3|Kmr2XqfMByWY`}Ain8XUMpV@cc{BuKCQjep58y&Oz-b`>4lQ~bIMxhbqP1tdWhBpZ-&Yty~Z^S$9u83=`Zb5ms$hmUlga_V=gHD-A?%#qrtW`$$zHAfca1<>05h-71hwl*-jE~~