Spaces:

bep40
/

ml-intern

Running

App Files Files Community

Auto-continue feature for free models

by bep40 - opened Jul 25

base: refs/heads/main

←

from: refs/pr/2

Discussion Files changed

+169

-8536

initial commit1bea302f

6 hidden commits · view all commits

Fix: Override model to use Anthropic API directly (not Bedrock)d468977f

Fix Dockerfile: copy local config override after clone43a72fd2

Use Qwen3-235B via HF Inference (free with Pro) - supports tool calling5ffc9159

Fix: Use novita provider for Qwen (works with standard HF token)34fe3cf6

Switch to OpenAI GPT-4o94bc0e74

Switch to Claude via OpenRouter6ea6949d

Switch to Claude Opus 4.8 via OpenRouter3a40ba5d

Fix: correct OpenRouter model ID for Claude Opus 4f3334f94

Fix: Use Claude Opus via OpenRouter by setting Anthropic base URL to OpenRouter77bf84b3

Fix: Use OpenAI format with OpenRouter base URL for Claude Opus 4e790ef50

Fix: Use HF router format with OpenRouter credentials passed as HF token trick935cb6f1

Fix: correct OpenRouter model name formata93741f6

Switch to DeepSeek V3 free via OpenRouter (has tool calling)0ee48339

Fix: correct model ID format for OpenRouter free model4bac0e2c

Switch to Gemini 2.5 Flash (cheap, fast, good tool calling)cfeccaa0

Try simpler model name format for OpenRouterfc4e6a18

Fix: correct Gemini 2.5 Flash model ID on OpenRoutereacab801

Fix: Override available models to use OpenRouter free models + correct model IDs0e22576c

Fix: Try direct deepseek model without openai prefix - let llm_params handle HF router formate8417235

Add llm_params patch for OpenRouter support + fix Dockerfile to apply it9edbd2e4

Fix Dockerfile: apply OpenRouter patch + custom start script0057d5b6

Add patched start script that loads OpenRouter supporte9afd249

Fix: start script that patches in-process before uvicorna4050b39

Override backend/main.py with OpenRouter patch at startupabcc4f07

Dockerfile: copy our backend/main.py override after source clone1ab656bc

Force rebuild: bust Docker cache + ensure overrides applye15170a9

ROOT FIX: Override agent.py with correct OpenRouter models + routingb53ebf87

ROOT FIX: Dockerfile patches agent.py models + llm_params at build time via sed8f678676

Fix: Use Python script to patch agent.py safely (no sed syntax issues)3c0610be

Add Python patch script for agent.py (safe replacement of models list)5a1d00b2

REAL FIX: Keep old model IDs in available list (frontend hardcodes them) but route ALL through OpenRouter7840d1c9

Fix main.py: Route ALL model IDs through OpenRouter regardless of prefixcb850a0e

CRITICAL: Also patch generate_title to use OpenRouter + fix title blocking chat74a225c1

Remove all custom files to restore 4bac0e28ad01c88

Restore Dockerfile from 4bac0e27ebd0531

Restore config from 4bac0e28d59c012

FINAL FIX: Patch frontend model list at Docker build time + route all via OpenRouter1a4d1637

Restore 4bac0e2 + add Gemini/Qwen3 to dropdown via build-time patchb3cb333a

Patch script: add DeepSeek/Gemini/Qwen3 to dropdown + fix title gen6b00697a

Fix patch: ensure models are added correctly to the list48b9248f

FIX: Patch frontend source to add Gemini+Qwen3 to dropdown BEFORE npm buildc79d1e94

Use Python script to patch frontend safely instead of sed5d14a4d5

Add frontend patch script to inject Gemini+Qwen3 into ChatInput.tsx62ab37db

Remove Kimi/MiniMax/GLM from frontend dropdownb74d1a54

Update frontend: Claude→DeepSeek V3, add Qwen3-235B HF Inference5556f90e

Backend patch: accept all new model IDs + route Qwen3-235B via HF Inference79682bef

Fix patch_frontend.py - use correct pattern from actual source317f7fae

Restore 5556f90 + add Grok4.3/DeepSeekV4Pro/Qwen3.7Max/Qwen3-HF to dropdown0c286f81

Update backend patch: accept all new model IDs + HF Inference for Qwen3f15d8ed0

Restore Dockerfile from 5556f90 (unchanged)123cd2ec

Fix frontend patch: rename Claude→DeepSeekV3, remove Kimi K2.6f1cdc24b

Fix patch_frontend.py - handle edge cases in string search4b77e8d6

Simplify: only ADD models, dont remove/edit existing entries (safest approach)ef0c5718

Restore 0c286f8: Dockerfile402e315b

Restore 0c286f8: patch_frontend.pyad34d68e

Restore 0c286f8: patch_models.py0404053a

Restore to f15d8ed541530a0

Fix models: DeepSeek V3 via OpenRouter as default + fix model IDs05bb9715

Fix model_switcher: rename Claude Opus 4.6 to DeepSeek V3, fix Qwen3/Grok/DeepSeek IDs for OpenRouter90d7b125

Fix: Dockerfile copies local overrides AFTER clone + correct model format0df23d6c

Fix: Use deepseek/ provider directly (litellm native support, no openrouter prefix needed)e9601f8f

Delete agent/f4ce4056

Restore 0c286f8: Dockerfile73d7d5fd

Restore 0c286f8: configs + patchesc3b14ebd

Update model IDs: Qwen3-Coder-Next, MiniMax-M2.7, GLM-5.1, DeepSeek-V4-Pro88430b6d

Update patch_models.py with correct HF Inference model IDs308b5f5f

Update default model to DeepSeek-V4-Proa00dc3f0

Replace Grok 4.3 with Owl Alpha (OpenRouter)928616d7

Replace Grok 4.3 with owl-alpha in backend models1fd2dc23

Fix title generation: add huggingface/ prefix for litellm520acbc7

Fix title generation: use openrouter/owl-alpha for title3955f857

Set default model to openrouter/owl-alpha0766391a

Set DEFAULT_FREE_MODEL_ID to openrouter/owl-alphaf5ef2f52

Fix: correct litellm model ID for OpenRouter = openrouter/openrouter/owl-alpha35310615

Fix model IDs: openrouter/openrouter/owl-alpha format for litellm4aca91ca

Fix frontend model path to openrouter/openrouter/owl-alphaf2597301

Try openai/ prefix with api_base for OpenRouter owl-alphae9905a1d

Fix: use openai/ prefix + api_base for OpenRouter models in litellm6f63bf1a

Fix frontend modelPath for owl-alpha67bba553

Fix: DeepSeek V4 Pro via OpenRouter + rename Claude to Owl Alpha5585e5dd

Frontend: DeepSeek via OpenRouter + rename Claude to Owl Alpha92ae1467

Update all model IDs as requested43495a6a

Frontend: update model list3d39bbc8

Fix: patch existing dropdown items exactly as specified64fe76f5

Fix frontend: patch existing dropdown entries exactly9e3b9509

Config default to deepseek-ai/DeepSeek-V4-Procad21448

Restore 3d39bbc + change: Kimi→DeepSeek-V4-Pro, MiniMax→deepseek-v4-flash(OR), GLM→DeepSeek-V4-Flash(HF)e38233f4

Restore 3d39bbc: frontend config06582f65

Restore 3d39bbc: patch_frontend.py (unchanged)34932a26

Fix: patch_frontend also renames existing models in dropdown (Kimi→DeepSeek, MiniMax→Flash, GLM→Flash)dfa59758

Fix patch: kimi→DeepSeek-V4-Pro(HF), minimax→deepseek-v4-flash(OR), glm→DeepSeek-V4-Flash(HF)550e63c5

Fix patch_models: replace kimi/minimax/glm with DeepSeek models in backend9a844a39

Fix model IDs: kimi→DeepSeek-V4-Pro, minimax→DeepSeek-V4-Flash(HF), glm→deepseek-v4-flash(OR)eba61d53

Update backend models to match frontend fixes41506713

Rename Space + update model dropdown names with performance notes315c11a6

Update dropdown names with performance notes5743f668

Fix: model_name format for litellm OpenRouter = openrouter/owl-alpha3bcc69a6

Fix: Space title + Owl Alpha display name + openai/ prefix for OpenRouterd063e525

Fix Owl Alpha: name display + openai/ prefix for OpenRouter routinge06f11b4

Fix backend: owl-alpha with openai/ prefixeaebe20f

Fix: litellm format openrouter/openrouter/owl-alpha + force display nameadf3016e

Fix backend: litellm openrouter/openrouter/owl-alpha + set as default5317d386

Fix config default model to openrouter format43ab5ae9

Fix: Owl Alpha display name + correct model paths with openai/ prefixdceca0a3

Fix: patch_models with openai/ prefix for OpenRouter models (owl-alpha, gemini-flash)2e719ba5

Fix: default model config2c838b31

Restore to 5743f66 + rename Claude Opus 4.6 → Owl Alpha1540d816

Restore configs from 5743f66b95ef5cf

Restore patch_models.py from 5743f660ef7b615

Restore patch_frontend.py + rename Claude Opus 4.6 → Owl Alphafea82a5d

Fix: patch both single+double quotes for Claude rename + block API label overridedf067833

Fix dropdown: rename providers + change model IDs for rows 4,5,6,725c9526a

Update backend patch: add nvidia/nemotron + google/gemma model IDs90073ea1

Fix dropdown names: row 4 Novita→HF Inference, row 6 name→nvidia/nemotron path5203fb14

Trigger rebuild to recover Space error8424baa6

Fix Space startup: use sh instead of missing bash9e7bfd9e

Add Space error fix report4aac713c

Upload patch_models.py36e2e7f8

Upload patch_frontend.py112ea82b

Upload patch_models.py3de92a80

Upload patch_frontend.pyb858c707

Upload patch_frontend.pyc8470eb6

Upload patch_frontend.py409739ff

Delete patch_frontend.py, patch_models.py, fix_space_error_report.json, configs/frontend_agent_config.json1d342460

Upload Dockerfile2ef1d92b

Upload Dockerfile59cbcf0a

Upload configs/frontend_agent_config.jsonbf69d068

Upload configs/cli_agent_config.jsonaa70a336

Upload configs/__init__.py8740b02a

Upload README.md8d69e4a5

Restore to commit 36e2e7f8 — remove extra config files7deaeea7

Restore fix_space_error_report.json from commit 36e2e7f81eb9d0c5

Restore patch_models.py from commit 36e2e7f833b84e4e

Restore Dockerfile from commit 36e2e7f8e3310d6d

Restore patch_frontend.py from commit 36e2e7f8de5a3788

Restore README.md from commit 36e2e7f8471f5b14

Restore frontend_agent_config.json from commit 36e2e7f85b9576b3

Upload fix_space_error_report.jsonc88a0e0a

Upload patch_frontend.pyd4b57ad3

Upload Dockerfiledc49b690

Upload README.mdc1bee844

Upload patch_models.py941d2b6a

Upload patch_models.py786a4bd5

Upload patch_frontend.pyf1d70ad8

Upload README.mdc59285d0

Upload README.md402458c0

Upload Dockerfile8d4088ce

Upload patch_frontend.pyeae5dafd

Upload README.mdb7568a8d

Remove Owl Alpha, change hy3 to tencent/hy3, make it default513dcd74

Change default model to tencent/hy322570ba6

Upload patch_models.pyfe500f29

Upload patch_frontend.pyced7c6bf

Upload configs/frontend_agent_config.jsonc7f974dc

Upload README.md0ff1a793

Upload README.mdd2c020b6

Upload patch_frontend.py1de664ba

Upload patch_models.pyc955b524

Upload configs/frontend_agent_config.jsond2a5abde

Upload README.md801e31c5

Upload configs/frontend_agent_config.jsonc7814487

Upload patch_frontend.pyd988d4d8

Upload patch_models.py6e160603

Upload configs/frontend_agent_config.json0d58290f

Upload patch_models.pyd2c26d16

Upload README.mdda9172cd

Upload patch_frontend.py00252799

Upload configs/frontend_agent_config.json69e60801

Upload patch_models.py8d9083a4

Upload patch_frontend.pyc1d2b498

Upload README.md51955633

Upload README.mdc52654fc

Upload patch_models.py35754bea

Upload configs/frontend_agent_config.json56031415

Upload patch_frontend.pyc74e807b

Upload patch_frontend.pye9e0d7af

Upload README.mdef9206b8

Upload configs/frontend_agent_config.json2b9f407e

Upload patch_models.py98d3509e

Upload configs/frontend_agent_config.json76ae8bc5

Upload patch_models.py53ddae08

Upload README.md5e3cda17

Upload patch_frontend.py6d199101

Add LAGUNA_S21_FREE_MODEL_ID, replace GPT-5.5 → Laguna S 2.1:free as DEFAULT_GPT_MODEL_ID38703994

Update README: replace GPT-5.5 with Laguna S 2.1:free, trigger rebuilda195002c

Add Laguna S 2.1:free to frontend dropdown, replace GPT-5.5f2a46251

Upload AUTO_CONTINUE_FEATURE.mdbf6e039b

Upload patch_auto_continue.py49755265

Upload patch_chat_input.py274c55b2

Upload patch_use_agent_chat_full.py90bab837

Upload patch_agent_loop.pye538ea2e

Upload Dockerfilee29e453a

Update Dockerfile: add auto-continue patches for free modelsaa15811d

Upload patch_chat_input.py769c39be

Upload patch_use_agent_chat_full.pyeb397197

bep40

Owner Jul 25

🤖

bep40/Space-Mutiverse-Control

Add auto-continue (10s) and pause buttons when free models stop mid-task

bep40 changed pull request status to open Jul 25

Add auto-continue feature for free models7694ee50

Add ChatInput patch for auto-continue38775a4d

Add SSE transport patch for task_incomplete event5dd8d788

Add backend patch for auto-continue detection98f080be

Update Dockerfile with all auto-continue patches836ae39f

bep40 changed pull request status to merged Jul 25

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment