<h1 id="flux-model-training-summary-(lora)">⚡ Flux Model Training Summary (LoRA)</h1>This job customized the Flux base model, which is a different architecture from Stable Diffusion XL. This run featured several key changes in its configuration, resulting in a longer duration.<h2 id="1.-job-status-and-timeline">1. Job Status and Timeline</h2><ul><li>Base Model Used: Flux (A modern, likely newer or different image generation model).</li><li>Job ID: 563759-20251018091302678</li><li>Training Start: October 18, 2025 at 05:13:06 PM</li><li>Completion Time (Ready): October 19, 2025 at 12:03:14 PM</li><li>Total Duration: Approximately 18 hours and 50 minutes (significantly longer than the previous job).</li><li>Dataset Used: 27 Files / 27 Labels.</li><li>Label Type: Caption (using full descriptive sentences, rather than just short tags, which is ideal for a complex model like Flux).</li></ul><h2 id="2.-key-training-parameters-(the-differences)">2. Key Training Parameters (The Differences)</h2>This configuration is notably different from the previous SDXL job, suggesting an attempt to train a very lightweight, highly specialized LoRA.<h3 id="model-intensity-and-focus">Model Intensity &amp; Focus</h3><ul><li>Network Dimension (networkDim): 2 (This is extremely low compared to 32 in the last job. This makes the resulting LoRA file much smaller but potentially less capable of capturing complex details).</li><li>Network Alpha (networkAlpha): 16 (Used for LoRA stabilization).</li><li>LoRA Type: lora.</li><li>Engine: kohya.</li></ul><h3 id="learning-process">Learning Process</h3><ul><li>Image Resolution: 512 (Lower resolution than the previous 1024).</li><li>Maximum Epochs: 35 (Higher than the previous 20, meaning the data was shown to the model more often to compensate for the small network size).</li><li>Text Encoder Learning Rate (textEncoderLR): 0 (This is crucial: The text understanding part of the model was frozen and did not learn. Only the image generation part (U-Net) was updated).</li><li>U-Net Learning Rate (unetLR): 0.0005.</li><li>Optimizer Type: AdamW8Bit (Different from Adafactor).</li><li>Shuffle Caption: false (Captions were fed in the same order each time).</li><li>Keep Tokens: 0 (No specific caption words were prioritized, unlike 3 previously).</li></ul>

2025-10-18-old-man-fit

Training Tags Exhaustive List (Chweee1 Focus)This is the complete, consolidated list of all tags used to train the LoRA model based on the second dataset.Subject &amp; Physical Attributes<ul><li>Subject: chweee1, male, Asian, muscular man, artist, everyday hero, mature 60 year old daddy</li><li>Physique/Body: muscular, medium build, muscular arms, abs, defined abs, six-pack, flexing bicep, tan skin, light skin, arm hair, chest hair, slight hair on stomach, visible arm veins, natural body</li><li>Tattoos &amp; Grooming: palm tree tattoo on left shoulder, tribal tattoo on right shoulder, visible pubic hair, slight tan lines, clean-shaven, unshaven, casual haircut</li><li>Facial Features: short black hair, glasses, slight smile, relaxed expression, confident pose, neutral expression</li></ul>Clothing &amp; Accessories<ul><li>Casual Tops: dark gray t-shirt, black t-shirt, grey t-shirt, black and white striped shirt, sleeveless gray tank top, grey polo shirt</li><li>Casual Bottoms: gray camouflage pants, blue camouflage pants, grey camo shorts, gray sweatpants, beige pants, black pants</li><li>NSFW/Underwear: naked, shirtless, blue boxer briefs, blue underwear, white drawstring</li><li>Accessories: black smartwatch, pocket on t-shirt</li></ul>NSFW Content Details<ul><li>Explicit: erect cock, large cock, medium-sized cock, hand on cock, uncircumcised tip, visible veins on cock, flaccid foreskin</li><li>Contextual: nsfw, lifting shirt, exposing abs, showing abs, low-angle shot, close-up, focus on torso and cock</li></ul>Environments &amp; Settings<ul><li>Home: bedroom, living room, white walls, bed with white pillows, gray headboard, wooden stool, wooden coffee table, TV, gray leather couch</li><li>Work/Studio: office setting, art studio, white desk, office cubicles, cityscape background, framed art, air conditioning unit, computer monitors</li><li>Art/Decor: abstract painting, colorful abstract painting, paintings on wall, paint palette, brushes, mixed media</li><li>Lighting/Composition: photograph, sfw, medium close-up, eye-level shot, rule of thirds, natural light, soft lighting, sunlight, strong shadows</li></ul>

Files with SHA256: 68b4226af592ddb326697b7fce23562f6463df83c2921551ac3ae0b6a425b6de