lemonade-sdk 8.1.7__py3-none-any.whl → 8.1.8__py3-none-any.whl

{lemonade_sdk-8.1.7.dist-info → lemonade_sdk-8.1.8.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lemonade-sdk
-Version: 8.1.7
+Version: 8.1.8
 Summary: Lemonade SDK: Your LLM Aide for Validation and Deployment
 Author-email: lemonade@amd.com
 Requires-Python: >=3.10, <3.14
@@ -45,42 +45,6 @@ Requires-Dist: human-eval-windows==1.0.4; extra == "dev"
 Requires-Dist: lm-eval[api]; extra == "dev"
 Provides-Extra: model-generate
 Requires-Dist: model-generate==1.5.0; (platform_system == "Windows" and python_version == "3.10") and extra == "model-generate"
-Provides-Extra: oga-hybrid
-Requires-Dist: lemonade-sdk[oga-ryzenai]; extra == "oga-hybrid"
-Provides-Extra: oga-unified
-Requires-Dist: lemonade-sdk[oga-ryzenai]; extra == "oga-unified"
-Provides-Extra: oga-hybrid-minimal
-Requires-Dist: lemonade-sdk[oga-ryzenai]; extra == "oga-hybrid-minimal"
-Provides-Extra: oga-cpu-minimal
-Requires-Dist: lemonade-sdk[oga-cpu]; extra == "oga-cpu-minimal"
-Provides-Extra: oga-npu-minimal
-Requires-Dist: lemonade-sdk[oga-ryzenai]; extra == "oga-npu-minimal"
-Provides-Extra: llm
-Requires-Dist: lemonade-sdk[dev]; extra == "llm"
-Provides-Extra: llm-oga-cpu
-Requires-Dist: lemonade-sdk[dev,oga-cpu]; extra == "llm-oga-cpu"
-Provides-Extra: llm-oga-npu
-Requires-Dist: onnx==1.16.0; extra == "llm-oga-npu"
-Requires-Dist: onnxruntime==1.18.0; extra == "llm-oga-npu"
-Requires-Dist: numpy==1.26.4; extra == "llm-oga-npu"
-Requires-Dist: protobuf>=6.30.1; extra == "llm-oga-npu"
-Requires-Dist: lemonade-sdk[dev]; extra == "llm-oga-npu"
-Provides-Extra: llm-oga-hybrid
-Requires-Dist: onnx==1.16.1; extra == "llm-oga-hybrid"
-Requires-Dist: numpy==1.26.4; extra == "llm-oga-hybrid"
-Requires-Dist: protobuf>=6.30.1; extra == "llm-oga-hybrid"
-Provides-Extra: llm-oga-unified
-Requires-Dist: lemonade-sdk[dev,llm-oga-hybrid]; extra == "llm-oga-unified"
-Provides-Extra: llm-oga-igpu
-Requires-Dist: onnxruntime-genai-directml==0.6.0; extra == "llm-oga-igpu"
-Requires-Dist: onnxruntime-directml<1.22.0,>=1.19.0; extra == "llm-oga-igpu"
-Requires-Dist: transformers<4.45.0; extra == "llm-oga-igpu"
-Requires-Dist: lemonade-sdk[dev]; extra == "llm-oga-igpu"
-Provides-Extra: llm-oga-cuda
-Requires-Dist: onnxruntime-genai-cuda==0.8.2; extra == "llm-oga-cuda"
-Requires-Dist: onnxruntime-gpu>=1.22.0; extra == "llm-oga-cuda"
-Requires-Dist: transformers<=4.51.3; extra == "llm-oga-cuda"
-Requires-Dist: lemonade-sdk[dev]; extra == "llm-oga-cuda"
 Dynamic: author-email
 Dynamic: description
 Dynamic: description-content-type
@@ -94,41 +58,29 @@ Dynamic: summary
 
 <p align="center">
   <a href="https://discord.gg/5xXzkMu8Zk">
-    <img src="https://img.shields.io/badge/Discord-7289DA?logo=discord&logoColor=white" alt="Discord" />
-  </a>
+    <img src="https://img.shields.io/badge/Discord-7289DA?logo=discord&logoColor=white" alt="Discord" /></a>
   <a href="https://github.com/lemonade-sdk/lemonade/tree/main/test" title="Check out our tests">
-    <img src="https://github.com/lemonade-sdk/lemonade/actions/workflows/test_lemonade.yml/badge.svg" alt="Lemonade tests" />
-  </a>
+    <img src="https://github.com/lemonade-sdk/lemonade/actions/workflows/test_lemonade.yml/badge.svg" alt="Lemonade tests" /></a>
   <a href="docs/README.md#installation" title="Check out our instructions">
-    <img src="https://img.shields.io/badge/Windows-11-0078D6?logo=windows&logoColor=white" alt="Windows 11" />
-  </a>
+    <img src="https://img.shields.io/badge/Windows-11-0078D6?logo=windows&logoColor=white" alt="Windows 11" /></a>
   <a href="https://lemonade-server.ai/#linux" title="Ubuntu 24.04 & 25.04 Supported">
-    <img src="https://img.shields.io/badge/Ubuntu-24.04%20%7C%2025.04-E95420?logo=ubuntu&logoColor=white" alt="Ubuntu 24.04 | 25.04" />
-  </a>
+    <img src="https://img.shields.io/badge/Ubuntu-24.04%20%7C%2025.04-E95420?logo=ubuntu&logoColor=white" alt="Ubuntu 24.04 | 25.04" /></a>
   <a href="docs/README.md#installation" title="Check out our instructions">
-    <img src="https://img.shields.io/badge/Python-3.10--3.13-blue?logo=python&logoColor=white" alt="Made with Python" />
-  </a>
+    <img src="https://img.shields.io/badge/Python-3.10--3.13-blue?logo=python&logoColor=white" alt="Made with Python" /></a>
   <a href="https://github.com/lemonade-sdk/lemonade/blob/main/docs/contribute.md" title="Contribution Guide">
-    <img src="https://img.shields.io/badge/PRs-welcome-brightgreen.svg" alt="PRs Welcome" />
-  </a>
+    <img src="https://img.shields.io/badge/PRs-welcome-brightgreen.svg" alt="PRs Welcome" /></a>
   <a href="https://github.com/lemonade-sdk/lemonade/releases/latest" title="Download the latest release">
-    <img src="https://img.shields.io/github/v/release/lemonade-sdk/lemonade?include_prereleases" alt="Latest Release" />
-  </a>
+    <img src="https://img.shields.io/github/v/release/lemonade-sdk/lemonade?include_prereleases" alt="Latest Release" /></a>
   <a href="https://tooomm.github.io/github-release-stats/?username=lemonade-sdk&repository=lemonade">
-    <img src="https://img.shields.io/github/downloads/lemonade-sdk/lemonade/total.svg" alt="GitHub downloads" />
-  </a>
+    <img src="https://img.shields.io/github/downloads/lemonade-sdk/lemonade/total.svg" alt="GitHub downloads" /></a>
   <a href="https://github.com/lemonade-sdk/lemonade/issues">
-    <img src="https://img.shields.io/github/issues/lemonade-sdk/lemonade" alt="GitHub issues" />
-  </a>
+    <img src="https://img.shields.io/github/issues/lemonade-sdk/lemonade" alt="GitHub issues" /></a>
   <a href="https://github.com/lemonade-sdk/lemonade/blob/main/LICENSE">
-    <img src="https://img.shields.io/badge/License-Apache-yellow.svg" alt="License: Apache" />
-  </a>
+    <img src="https://img.shields.io/badge/License-Apache-yellow.svg" alt="License: Apache" /></a>
   <a href="https://github.com/psf/black">
-    <img src="https://img.shields.io/badge/code%20style-black-000000.svg" alt="Code style: black" />
-  </a>
+    <img src="https://img.shields.io/badge/code%20style-black-000000.svg" alt="Code style: black" /></a>
   <a href="https://star-history.com/#lemonade-sdk/lemonade">
-    <img src="https://img.shields.io/badge/Star%20History-View-brightgreen" alt="Star History Chart" />
-  </a>
+    <img src="https://img.shields.io/badge/Star%20History-View-brightgreen" alt="Star History Chart" /></a>
 </p>
 <p align="center">
   <img src="https://github.com/lemonade-sdk/assets/blob/main/docs/banner.png?raw=true" alt="Lemonade Banner" />
@@ -156,7 +108,7 @@ Startups such as [Styrk AI](https://styrk.ai/styrk-ai-and-amd-guardrails-for-you
 ### Use it with your favorite OpenAI-compatible app!
 
 <p align="center">
-  <a href="https://lemonade-server.ai/docs/server/apps/open-webui/" title="Open WebUI" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/openwebui.jpg" alt="Open WebUI" width="60" /></a>&nbsp;&nbsp;<a href="https://lemonade-server.ai/docs/server/apps/continue/" title="Continue" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/continue_dev.png" alt="Continue" width="60" /></a>&nbsp;&nbsp;<a href="https://github.com/amd/gaia" title="Gaia" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/gaia.ico" alt="Gaia" width="60" /></a>&nbsp;&nbsp;<a href="https://lemonade-server.ai/docs/server/apps/anythingLLM/" title="AnythingLLM" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/anything_llm.png" alt="AnythingLLM" width="60" /></a>&nbsp;&nbsp;<a href="https://lemonade-server.ai/docs/server/apps/ai-dev-gallery/" title="AI Dev Gallery" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/ai_dev_gallery.webp" alt="AI Dev Gallery" width="60" /></a>&nbsp;&nbsp;<a href="https://lemonade-server.ai/docs/server/apps/lm-eval/" title="LM-Eval" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/lm_eval.png" alt="LM-Eval" width="60" /></a>&nbsp;&nbsp;<a href="https://lemonade-server.ai/docs/server/apps/codeGPT/" title="CodeGPT" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/codegpt.jpg" alt="CodeGPT" width="60" /></a>&nbsp;&nbsp;<a href="https://github.com/lemonade-sdk/lemonade/blob/main/docs/server/apps/ai-toolkit.md" title="AI Toolkit" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/ai_toolkit.png" alt="AI Toolkit" width="60" /></a>
+  <a href="https://lemonade-server.ai/docs/server/apps/open-webui/" title="Open WebUI" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/openwebui.jpg" alt="Open WebUI" width="60" /></a>&nbsp;&nbsp;<a href="https://lemonade-server.ai/docs/server/apps/continue/" title="Continue" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/continue_dev.png" alt="Continue" width="60" /></a>&nbsp;&nbsp;<a href="https://github.com/amd/gaia" title="Gaia" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/gaia.ico" alt="Gaia" width="60" /></a>&nbsp;&nbsp;<a href="https://lemonade-server.ai/docs/server/apps/anythingLLM/" title="AnythingLLM" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/anything_llm.png" alt="AnythingLLM" width="60" /></a>&nbsp;&nbsp;<a href="https://lemonade-server.ai/docs/server/apps/ai-dev-gallery/" title="AI Dev Gallery" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/ai_dev_gallery.webp" alt="AI Dev Gallery" width="60" /></a>&nbsp;&nbsp;<a href="https://lemonade-server.ai/docs/server/apps/lm-eval/" title="LM-Eval" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/lm_eval.png" alt="LM-Eval" width="60" /></a>&nbsp;&nbsp;<a href="https://github.com/lemonade-sdk/lemonade-arcade" title="Lemonade Arcade" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/lemonade-arcade/refs/heads/main/docs/assets/favicon.ico" alt="Lemonade Arcade" width="60" /></a>&nbsp;&nbsp;<a href="https://github.com/lemonade-sdk/lemonade/blob/main/docs/server/apps/ai-toolkit.md" title="AI Toolkit" target="_blank"><img src="https://raw.githubusercontent.com/lemonade-sdk/assets/refs/heads/main/partner_logos/ai_toolkit.png" alt="AI Toolkit" width="60" /></a>
 </p>
 
 > [!TIP]
@@ -227,7 +179,7 @@ Lemonade supports the following configurations, while also making it easy to swi
     </tr>
     <tr>
       <td><b>gfx120X</b> (RDNA4)</td>
-      <td>Windows only</td>
+      <td>Windows, Ubuntu</td>
       <td>Radeon AI PRO R9700, RX 9070 XT/GRE/9070, RX 9060 XT</td>
     </tr>
     <tr>

{lemonade_sdk-8.1.7.dist-info → lemonade_sdk-8.1.8.dist-info}/RECORD

@@ -1,10 +1,10 @@
 lemonade/__init__.py,sha256=W1Qk7r0rnQqFhPNHp6BIBT_q-OH3s-8Q_POoVfAmKW0,117
 lemonade/api.py,sha256=Oc4yBA3LZg8FrTsbuDq1p9-XE74pqNnIEUhXyKa7qg8,5786
 lemonade/cache.py,sha256=5iZbk273TiTMqK_vdzPOPYTo6VsWW2gNByOISA9zi1w,3002
-lemonade/cli.py,sha256=9Pcs3PcrWC2F8_pcBaz09xHUICIJTvpemBdPGyXkjIk,4395
+lemonade/cli.py,sha256=2dWGg3u_iCS4zVfLImYhPEOBcw1_JqY2doCaQL6hYpM,4237
 lemonade/sequence.py,sha256=KSH7BPsiyDKsOsg_ziQKEGsDwMmuO_YbgPRBxkZd0pw,13267
 lemonade/state.py,sha256=sdSezla7Cd7KYL90xY3p9kcNV4ndSyN6UvNLOr3vBMA,5261
-lemonade/version.py,sha256=DSjpcTXEMbKTSEhH-ChqB5reU3Hdp38wfe3HkEnrtRQ,22
+lemonade/version.py,sha256=KEEcawgJAxFWpeDw-nZyFzFwfyl_0inM8UDk3-FNKWE,22
 lemonade/common/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 lemonade/common/build.py,sha256=zTb0m1-kuUx6zw5QHp2SNnVuN6jOTMQ2FCdj9iH374U,6140
 lemonade/common/cli_helpers.py,sha256=hjBfXrTtFl8gmCFlL-ksviXR0mOcdPtTWVNKoEp3PG4,4993
@@ -39,40 +39,37 @@ lemonade/tools/oga/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuF
 lemonade/tools/oga/bench.py,sha256=PJXv4UchcS2YPwijNzef8DY4DSAKYxIYY1ycHuH3T34,5005
 lemonade/tools/oga/load.py,sha256=BH5ChYbZgeP_ZN4E6HoboJD3kZcUIAPgPEVbgUZpVjQ,33778
 lemonade/tools/oga/utils.py,sha256=F8UVLKlfYcLa2SUqlehar8-jaX2Aw4u58DjHNNvLdOA,17675
-lemonade/tools/quark/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-lemonade/tools/quark/quark_load.py,sha256=FJ4LJKTToZbHHWVEOBLadae1a3jCnnY4KvXySHbkJMA,5589
-lemonade/tools/quark/quark_quantize.py,sha256=hwoaXhpBIORvJ16MvewphPkaDEQn3BAgXq5o82Gc-_s,16599
 lemonade/tools/report/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 lemonade/tools/report/llm_report.py,sha256=bVHhwCINA-Ok2EdSwAsLubsc83N3KWOVuwTguw7jDcE,6676
-lemonade/tools/report/table.py,sha256=ssqy1bZqF-wptNzKEOj6_9REtCNZyXO8R5vakAtg3R4,27973
+lemonade/tools/report/table.py,sha256=Kv_Epd8a6KIrdzSC2EgIl6uTKw7E5eMq10Tg16O0WxM,27996
 lemonade/tools/server/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-lemonade/tools/server/llamacpp.py,sha256=w-M0JXrgXVpfICnHBHhJm_yBshfwZ8zge-e1o1kH0R4,8751
-lemonade/tools/server/serve.py,sha256=YVsO7m9E1OsQtbcjkv_1ir8oCSAdr0IXGNEvJ6FKamw,60311
+lemonade/tools/server/llamacpp.py,sha256=a_rO4ElrUbxlMeNx-8Ru6QQvgFhylotE-4-76bVAdr4,9358
+lemonade/tools/server/serve.py,sha256=gwKPs-r-bfQsyCkVxzVyTp0SyNYVqqfMeeUFQ9TivwM,61638
 lemonade/tools/server/tool_calls.py,sha256=xrAlQwKG-nv2xLlf8f9CDSaUbyMn8ZtHkds9iZLG9K8,5230
 lemonade/tools/server/tray.py,sha256=a9z6hdqlfj91H00j6hAExRPQkzWHhE3dnqSumzEgq0U,19599
 lemonade/tools/server/webapp.py,sha256=8Das5yXOaSBLZmSZ_eddJajQFxBhvl5D6GI_hHlGbE0,1040
-lemonade/tools/server/wrapped_server.py,sha256=-knOr2ycmrebVPYrOlCNRJH0ySZPnVlWzbKYLsfTRhE,16441
+lemonade/tools/server/wrapped_server.py,sha256=DlzsGUwLQzjOFRfTTxfnhvmM_9lvAki96jWIAz7Czds,16713
 lemonade/tools/server/static/favicon.ico,sha256=hMmP9qGJNeZ0mFS86JIqPbZstXMZn0Z76_HfHQpREAU,126745
-lemonade/tools/server/static/styles.css,sha256=5HQQCpm8N_fzLcolPiDuhyZw_5nbO8aIl60xAn4RKmg,43385
-lemonade/tools/server/static/webapp.html,sha256=FX2MZUsljfgxxuF12KBdgvNkso_z-sHewWc0SEGGcGM,18138
-lemonade/tools/server/static/js/chat.js,sha256=BTvREuEt0NrN8qhAuda5tTAoUN6tbsoukevA-zyTrwQ,27193
+lemonade/tools/server/static/styles.css,sha256=TivveWAEyWnEXGJX_j22432LLWdwMljpW_70WuwIVsk,43730
+lemonade/tools/server/static/webapp.html,sha256=KbYl8_q0GOXHiZTAv96O1e2RujfWMy6DLA4Wc5IbomY,18173
+lemonade/tools/server/static/js/chat.js,sha256=vTfwJODpxkNPGG3J0AhfIF1WbraWnHNmbCQOsS-Fyc4,29488
 lemonade/tools/server/static/js/model-settings.js,sha256=JXHeG7xVrRU181Hj7CZflERAi1Z6t-qwYFR4aH5nf5I,5820
-lemonade/tools/server/static/js/models.js,sha256=bbX7c8B59ioim86T3x9PFESvF8y3cHPYUO6nhc4SCDs,32500
-lemonade/tools/server/static/js/shared.js,sha256=4iqDNWiKEB7eYS4fdnTy-RwO_ksROrLYLmT2YSomG1M,17065
+lemonade/tools/server/static/js/models.js,sha256=2ueyUfn7f4UeAvha9vyDCEi3fDk0BxLlut1N0L2f2EE,32862
+lemonade/tools/server/static/js/shared.js,sha256=56dcVnAPiM7hKQwDVszlXCufCgzGiZyfNec9B-ScsHo,17356
 lemonade/tools/server/utils/port.py,sha256=J7-g-Aqygb50jNoHLhhRfBZVM-uhGlcB5-oYBAehvgw,2263
 lemonade/tools/server/utils/system_tray.py,sha256=b9lvNv9chJKQxvmH7qzAuUe6H9HsLu7pdHFqGlAJaL0,12654
 lemonade/tools/server/utils/thread.py,sha256=Z-PDzGcpgfN2qxTmtlROWqrUN0B2fXdPrqo_J10fR_w,2772
 lemonade_install/__init__.py,sha256=26zohKg2jgr_5y7tObduWMYQg8zCTWMZHL8lfi2zZVQ,40
-lemonade_install/install.py,sha256=onndA2a-ygyLtDfupI8JQFhU_XpK8McGZtGujFasXww,28304
-lemonade_sdk-8.1.7.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-lemonade_sdk-8.1.7.dist-info/licenses/NOTICE.md,sha256=RSca9LE5e6pvdWA_LXAUCcACIHPmINKqkRX-AVRqBGo,3499
-lemonade_server/cli.py,sha256=GsYMg095XMADHjWr1ytLYvbbrtSDPwQJEUn0gI-W99o,19108
-lemonade_server/model_manager.py,sha256=xHbHo14q9EbueY6rxijFKUQtf4hgod0w9gBuPw2mVdk,18065
+lemonade_install/install.py,sha256=Dow7kt-K9WI4PH15hBwkKtOxede3dAaOmH4I1y_P5H4,27008
+lemonade_sdk-8.1.8.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+lemonade_sdk-8.1.8.dist-info/licenses/NOTICE.md,sha256=RSca9LE5e6pvdWA_LXAUCcACIHPmINKqkRX-AVRqBGo,3499
+lemonade_server/cli.py,sha256=IsyWGDIeiN0MnIaBryLhArHhPj1HD52l6028dNHQCKc,19248
+lemonade_server/model_manager.py,sha256=V8QRf1nlh3wAFtUHoSF_JeAXeR7sfaZE1uTfppcIfcw,20492
 lemonade_server/pydantic_models.py,sha256=49MyOlb5feLUlKsGcI75tWaflWckrItqcSVkdCY4e3A,3269
 lemonade_server/server_models.json,sha256=DAdG4ebIt5Dy5MM3kmXn1pO0XbNMph1gdpzbacBDVuc,11664
 lemonade_server/settings.py,sha256=6nsmPLFJD-UokQDmlx9ZBYMbpnn48So_PuBGWP7Fmfg,1299
-lemonade_sdk-8.1.7.dist-info/METADATA,sha256=onbpcmKIyz0x7FmhWIzGFD0YnyOoOTUTCxtcxSno1gA,16852
-lemonade_sdk-8.1.7.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-lemonade_sdk-8.1.7.dist-info/entry_points.txt,sha256=7sRvpNhi1E7amnM7RZo57e8yFF9iA5uuRaIeJ1Xre6w,193
-lemonade_sdk-8.1.7.dist-info/top_level.txt,sha256=10ap5GNiPhalO4V50LRoxA1FqRT9g3Xkia6BITu880k,42
-lemonade_sdk-8.1.7.dist-info/RECORD,,
+lemonade_sdk-8.1.8.dist-info/METADATA,sha256=EoGUK5KNOw1lCefStfN47N-3mUIdJeNGTtpDIE63i0E,14992
+lemonade_sdk-8.1.8.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+lemonade_sdk-8.1.8.dist-info/entry_points.txt,sha256=7sRvpNhi1E7amnM7RZo57e8yFF9iA5uuRaIeJ1Xre6w,193
+lemonade_sdk-8.1.8.dist-info/top_level.txt,sha256=10ap5GNiPhalO4V50LRoxA1FqRT9g3Xkia6BITu880k,42
+lemonade_sdk-8.1.8.dist-info/RECORD,,

lemonade_server/cli.py

@@ -197,15 +197,15 @@ def pull(
         for model_name in model_names:
             payload = {"model_name": model_name}
 
-            if checkpoint and recipe:
-                # Add the parameters for registering a new model
-                payload["checkpoint"] = checkpoint
-                payload["recipe"] = recipe
-
-                if reasoning:
-                    payload["reasoning"] = reasoning
-                if mmproj:
-                    payload["mmproj"] = mmproj
+            # Add the parameters to the payload
+            for key, value in [
+                ("checkpoint", checkpoint),
+                ("recipe", recipe),
+                ("reasoning", reasoning),
+                ("mmproj", mmproj),
+            ]:
+                if value:
+                    payload[key] = value
 
             # Install the model
             pull_response = requests.post(f"{base_url}/pull", json=payload)
@@ -225,6 +225,9 @@ def pull(
             recipe=recipe,
             reasoning=reasoning,
             mmproj=mmproj,
+            # The pull command will download an upgraded model if available, even
+            # if we already have a local copy of the model
+            do_not_upgrade=False,
         )
 
 

lemonade_server/model_manager.py

@@ -201,6 +201,54 @@ class ModelManager:
                     reasoning=reasoning,
                 )
             else:
+                # Model is already registered - check if trying to register with different parameters
+                existing_model = self.supported_models[model]
+                existing_checkpoint = existing_model.get("checkpoint")
+                existing_recipe = existing_model.get("recipe")
+                existing_reasoning = "reasoning" in existing_model.get("labels", [])
+                existing_mmproj = existing_model.get("mmproj", "")
+
+                # Compare parameters
+                checkpoint_differs = checkpoint and checkpoint != existing_checkpoint
+                recipe_differs = recipe and recipe != existing_recipe
+                reasoning_differs = reasoning and reasoning != existing_reasoning
+                mmproj_differs = mmproj and mmproj != existing_mmproj
+
+                if (
+                    checkpoint_differs
+                    or recipe_differs
+                    or reasoning_differs
+                    or mmproj_differs
+                ):
+                    conflicts = []
+                    if checkpoint_differs:
+                        conflicts.append(
+                            f"checkpoint (existing: '{existing_checkpoint}', new: '{checkpoint}')"
+                        )
+                    if recipe_differs:
+                        conflicts.append(
+                            f"recipe (existing: '{existing_recipe}', new: '{recipe}')"
+                        )
+                    if reasoning_differs:
+                        conflicts.append(
+                            f"reasoning (existing: {existing_reasoning}, new: {reasoning})"
+                        )
+                    if mmproj_differs:
+                        conflicts.append(
+                            f"mmproj (existing: '{existing_mmproj}', new: '{mmproj}')"
+                        )
+
+                    conflict_details = ", ".join(conflicts)
+
+                    additional_suggestion = ""
+                    if model.startswith("user."):
+                        additional_suggestion = f" or delete the existing model first using `lemonade-server delete {model}`"
+
+                    raise ValueError(
+                        f"Model {model} is already registered with a different configuration. "
+                        f"Conflicting parameters: {conflict_details}. "
+                        f"Please use a different model name{additional_suggestion}."
+                    )
                 new_registration_model_config = None
 
             # Download the model

lemonade/tools/quark/quark_load.py

@@ -1,173 +0,0 @@
-import argparse
-import os
-import sys
-
-from lemonade.state import State
-from lemonade.tools import Tool
-import lemonade.common.printing as printing
-import lemonade.common.build as build
-from lemonade_install.install import DEFAULT_QUARK_DIR
-
-
-class QuarkLoad(Tool):
-    """
-    Load a model Quantized and exported using Quark.
-    Required Input State:
-        - state.model: Pretrained model instance to be quantized.
-        - state.tokenizer: Tokenizer instance from Hugging Face.
-    Output:
-        - state of the loaded model
-
-    See docs/dev_cli/quark.md for more details.
-    """
-
-    unique_name = "quark-load"
-
-    def __init__(self):
-        super().__init__(monitor_message="Load Quark Quantized model")
-
-    @staticmethod
-    def parser(add_help: bool = True) -> argparse.ArgumentParser:
-        parser = __class__.helpful_parser(
-            short_description="Load a quantized model using Quark",
-            add_help=add_help,
-        )
-
-        parser.add_argument(
-            "--quant-scheme",
-            type=str,
-            required=True,
-            default=None,
-            help="Supported quantization schemes in Quark",
-        )
-
-        parser.add_argument(
-            "--quant-algo",
-            type=str,
-            required=True,
-            default=None,
-            choices=["awq", "gptq", "autosmoothquant", None],
-            help="Supported quantization algorithms in Quark",
-        )
-
-        parser.add_argument(
-            "--torch-compile", action="store_true", help="Model torch compile"
-        )
-
-        parser.add_argument(
-            "--safetensors-model-reload",
-            action="store_true",
-            help="Safetensors model reload",
-        )
-
-        parser.add_argument(
-            "--safetensors-model-dir",
-            default=None,
-            help="Directory of safetensors model",
-        )
-
-        parser.add_argument(
-            "--params-load", action="store_true", help="Model parameters load"
-        )
-
-        parser.add_argument("--json-path", help="Specify the path of saved json file")
-
-        parser.add_argument(
-            "--safetensors-path",
-            default=None,
-            help="Specify the path of saved safetensors file",
-        )
-
-        return parser
-
-    def run(
-        self,
-        state: State,
-        quant_scheme: str,
-        quant_algo: str,
-        torch_compile: bool = False,
-        safetensors_model_reload: bool = False,
-        safetensors_model_dir: str = None,
-        params_load: bool = False,
-        json_path: str = None,
-        safetensors_path: str = None,
-    ) -> State:
-        """
-        Executes the QuarkLoad process.
-        Returns:
-            State: The updated state after loading the model.
-        Raises:
-            Exception: If an error occurs during the QuarkLoad process.
-        """
-
-        import torch
-
-        try:
-            if os.path.isdir(DEFAULT_QUARK_DIR):
-                quark_llm_path = os.path.join(
-                    DEFAULT_QUARK_DIR, "examples", "torch", "language_modeling"
-                )
-                sys.path.insert(0, quark_llm_path)
-            else:
-                raise FileNotFoundError(
-                    f"The directory {DEFAULT_QUARK_DIR} does not exist. \
-                        Please check your installation."
-                )
-
-            # Default load path specific to recipe
-            # This will NOT work
-            # The default path is now uniquely craeated with timestamp
-            # Default load path will not work. Need to pass explicit load path
-            model_export_path = os.path.join(
-                build.output_dir(state.cache_dir, state.build_name),
-                "exported_model",
-                quant_scheme,
-                quant_algo,
-            )
-
-            # Set default paths only if current values are None
-            if safetensors_model_dir is None:
-                safetensors_model_dir = model_export_path
-            if safetensors_path is None:
-                safetensors_path = os.path.join(model_export_path, "model.safetensors")
-            printing.log_info("Loading model ...")
-            if not params_load and not safetensors_model_reload:
-                raise ValueError(
-                    " Specify load format: 'params_load' or 'safetensors_model_reload'."
-                )
-
-            # Reload quantized model if specified
-            from quark.torch import load_params, import_model_info
-
-            if params_load:
-                printing.log_info(
-                    "Restoring quantized model from JSON/safetensors files"
-                )
-                model = load_params(
-                    model,
-                    json_path=json_path,
-                    safetensors_path=safetensors_path,
-                )
-            elif safetensors_model_reload:
-                printing.log_info(
-                    "Restoring quantized model from quark_safetensors files"
-                )
-                model = import_model_info(model, model_info_dir=safetensors_model_dir)
-
-            if torch_compile:
-                printing.log_info("torch.compile...")
-                model = torch.compile(model)
-
-            state.model = model
-            state.dtype = model.dtype
-
-            printing.log_info("Quark Load process completed.")
-
-        except Exception as e:
-            printing.log_error(f"An error occurred during the QuarkLoad process: {e}")
-            raise
-        return state
-
-
-# This file was originally licensed under Apache 2.0. It has been modified.
-# Modifications Copyright (c) 2025 AMD