Xoron-Dev-MultiMoe / streaming_state.json
Backup-bdg's picture
Update model weights after training (epoch 4, loss 5.8919)
7936332 verified
{
"epoch": 170,
"unique_samples": 150,
"total_yields": 300,
"dataset_positions": {
"WebSight": 386,
"ScienceQA": 364,
"InstructPix2Pix": 386,
"Flickr8k": 386,
"NewYorker": 386,
"Football": 6,
"MagicBrush": 386,
"WildChat": 500,
"Synth-ShellExecution": 500,
"Midjourney-Prompts": 200,
"Synth-KnowledgeCutoff": 550,
"Synth-GroundedResponse": 550,
"CodeParrot-Clean": 350,
"ShareGPT-Clean": 500,
"Synth-Issues": 350,
"Dolly-15k": 800,
"Conversation-Summarization": 800,
"Synth-ShellTimeout": 500,
"Synth-Docker": 500,
"Synth-Documents": 450,
"HumanEval-JavaScript": 164,
"OpenOrca": 800,
"Synth-MultiStepExecution": 500,
"Synth-Citation": 550,
"NoRobots": 800,
"Synth-LanguageSetup": 500,
"Function-Calling-ChatML": 500,
"Synth-CoT": 900,
"Python-Code-18k": 350,
"Code-Feedback": 350,
"HumanEval-CPP": 164,
"AgentInstruct": 195,
"SD-Prompts": 200,
"Synth-Diffs": 350,
"Golang-Coder": 350,
"Synth-ConfidenceLevel": 550,
"Synth-RepoContext": 350,
"HumanEval-Go": 164,
"Synth-SelfCorrection": 550,
"Synth-FactCheck": 550,
"Synth-Downloads": 500,
"Synth-RetrievalGrounded": 550,
"Synth-IDK": 550,
"Synth-APIGen": 500,
"Synth-PythonScripts": 500,
"Synth-Uncertainty": 550,
"HumanEval-Python": 164,
"Golang-QA-2k": 350,
"Synth-ShellErrors": 500,
"Synth-Jupyter": 500,
"Jupyter-Code": 350,
"Synth-Execution": 500,
"Synth-Monitoring": 500,
"Synth-DatabaseSetup": 500,
"HumanEval-Java": 164,
"Synth-AptInstall": 500,
"UltraChat": 500,
"Synth-DesktopSetup": 500,
"SD-Prompts-2M": 200,
"Synth-WebserverSetup": 500,
"Pythonic-Function-Calling": 500,
"Swift-Code-Edit": 10,
"Glaive-Code-Assistant": 500,
"File-Operations-Medium": 500,
"Swift-Code-RLVR": 350,
"Synth-SSHSetup": 500,
"HumanEval-Rust": 164,
"Synth-Commits": 350,
"Synth-FIM": 350,
"Synth-Debugging": 500,
"Tool-Calls-SingleTurn": 500,
"Tool-Calls-Multiturn": 500,
"OpenAssistant": 800,
"T2V-Sora-Preferences-2": 650,
"T2V-Human-Preferences": 650,
"Sora-Alignment-Likert": 198,
"Sora-Style-Likert": 198,
"I2V-Preference-Seedance": 198,
"WebVid-10M": 650,
"Sora-Physics-Likert": 198,
"TIP-I2V": 650,
"Pexels-I2V-350k": 650,
"SmolTalk-OpenHermes": 600,
"SmolTalk-All": 600,
"Cosmopedia-AutoMath": 600,
"OpenMathInstruct-1": 600,
"NuminaMath-CoT": 600,
"UltraData-Math-Conv": 600,
"Cosmopedia-KhanAcademy": 600,
"NuminaMath-TIR": 600,
"UltraData-Math-QA": 600,
"Cosmopedia-OpenStax": 600,
"MedMCQA": 850,
"Medical-Reasoning-SFT-Mega": 850,
"Medical-O1-Reasoning-EN": 850,
"OpenThoughts-114k": 350,
"Bespoke-Stratos-17k": 350,
"Synth-FileOps": 300,
"Synth-EditLines": 300,
"Agentic-CoT-Coding": 300
},
"modality_positions": {
"text": {
"WildChat": 500,
"Midjourney-Prompts": 200,
"CodeParrot-Clean": 350,
"ShareGPT-Clean": 500,
"Dolly-15k": 800,
"Conversation-Summarization": 800,
"HumanEval-JavaScript": 164,
"OpenOrca": 800,
"NoRobots": 800,
"Function-Calling-ChatML": 500,
"Python-Code-18k": 350,
"Code-Feedback": 350,
"HumanEval-CPP": 164,
"AgentInstruct": 195,
"SD-Prompts": 200,
"Golang-Coder": 350,
"HumanEval-Go": 164,
"Synth-APIGen": 500,
"HumanEval-Python": 164,
"Golang-QA-2k": 350,
"Jupyter-Code": 350,
"HumanEval-Java": 164,
"UltraChat": 500,
"SD-Prompts-2M": 200,
"Pythonic-Function-Calling": 500,
"Swift-Code-Edit": 10,
"Glaive-Code-Assistant": 500,
"Swift-Code-RLVR": 350,
"HumanEval-Rust": 164,
"Tool-Calls-SingleTurn": 500,
"Tool-Calls-Multiturn": 500,
"OpenAssistant": 800,
"SmolTalk-OpenHermes": 600,
"SmolTalk-All": 600,
"Cosmopedia-AutoMath": 600,
"OpenMathInstruct-1": 600,
"NuminaMath-CoT": 600,
"UltraData-Math-Conv": 600,
"Cosmopedia-KhanAcademy": 600,
"NuminaMath-TIR": 600,
"UltraData-Math-QA": 600,
"Cosmopedia-OpenStax": 600,
"MedMCQA": 850,
"Medical-Reasoning-SFT-Mega": 850,
"Medical-O1-Reasoning-EN": 850,
"Synth-SelfCorrection": 550,
"Synth-GroundedResponse": 550,
"Synth-IDK": 550,
"Synth-KnowledgeCutoff": 550,
"Synth-RetrievalGrounded": 550,
"Synth-FactCheck": 550,
"Synth-ConfidenceLevel": 550,
"Synth-Citation": 550,
"Synth-Uncertainty": 550,
"OpenThoughts-114k": 350,
"Bespoke-Stratos-17k": 350,
"Synth-CoT": 900,
"Synth-RepoContext": 350,
"Synth-Issues": 350,
"Synth-Commits": 350,
"Synth-FIM": 350,
"Synth-Diffs": 350,
"Synth-Monitoring": 500,
"Synth-FileOps": 300,
"Synth-Debugging": 500,
"Synth-Downloads": 500,
"Synth-ShellErrors": 500,
"Synth-DesktopSetup": 500,
"Synth-ShellExecution": 500,
"Synth-LanguageSetup": 500,
"Synth-DatabaseSetup": 500,
"Synth-MultiStepExecution": 500,
"Synth-Jupyter": 500,
"File-Operations-Medium": 500,
"Synth-ShellTimeout": 500,
"Synth-Docker": 500,
"Synth-SSHSetup": 500,
"Synth-EditLines": 300,
"Synth-AptInstall": 500,
"Synth-Execution": 500,
"Synth-PythonScripts": 500,
"Synth-WebserverSetup": 500,
"Agentic-CoT-Coding": 300
},
"image": {
"WebSight": 386,
"ScienceQA": 364,
"InstructPix2Pix": 386,
"Flickr8k": 386,
"NewYorker": 386,
"Football": 6,
"MagicBrush": 386
},
"video": {
"T2V-Sora-Preferences-2": 650,
"T2V-Human-Preferences": 650,
"Sora-Alignment-Likert": 198,
"Sora-Style-Likert": 198,
"I2V-Preference-Seedance": 198,
"WebVid-10M": 650,
"Sora-Physics-Likert": 198,
"TIP-I2V": 650,
"Pexels-I2V-350k": 650
},
"audio": {}
},
"modality_counts": {
"text": 150,
"image": 0,
"video": 0,
"audio": 0
},
"last_modality": null
}