Add jobs taxonomy,exact prompts,metadata and progress (16)

README.md

@@ -23,14 +23,14 @@ configs:
     path:
     - data/*.tar
     - job-data/*.tar
-- config_name: jobs
-  data_files:
-  - split: train
-    path: job-data/*.tar
 - config_name: base
   data_files:
   - split: train
     path: data/*.tar
+- config_name: jobs
+  data_files:
+  - split: train
+    path: job-data/*.tar
 dataset_info:
 - config_name: default
   features: &id001
@@ -312,12 +312,27 @@ dataset_info:
     dtype: string
   - name: __url__
     dtype: string
-- config_name: jobs
-  features: *id001
 - config_name: base
   features: *id001
+- config_name: jobs
+  features: *id001
 ---
 
+# Complete character collection and professional outfit plan
+
+**Planned total: 840 identities, 12,140 images.** The expression/special collection contains 7,500 planned images. A further **464 adult humans × 10 professional outfits = 4,640 sprites** reuse those identities and their neutral references; they do not add new personality IDs. See [all 100 jobs and the exact outfit allocation](JOB_OUTFITS.md). Job sprites uploaded so far: **16/4,640**. The `jobs` streaming configuration becomes available when its first TAR is uploaded.
+
+The [teen/child detector recheck](SAFETY_RECHECK.md) reports actual image checksums, both LAION classifier scores, operational thresholds and any replacements. Intended ages, professions and demographic attributes describe prompting, not inferred real-world classifications.
+
+# Vivarium warm-painted greenscreen people and special characters
+
+**Combined uploaded snapshot: 5,000/7,500 images.** The collection comprises an original 500-person series, an extension of 240 adult people with the same ten emotions, and 100 fantasy / science-fiction specials with one neutral portrait each. All images are 832×1,248, in the selected Ghibli-inspired warm-painted style, with solid greenscreen backgrounds. The original collection has 5,000/5,000 images uploaded; the extension has 0/2,500.
+
+See the detailed [extension allocation and 25-class special taxonomy](EXTENSIONS.md), including all 100 variants, exact prompt plans, CSVs, inference examples and measured progress. The base model is [FLUX.2-klein-base-9B](https://huggingface.co/black-forest-labs/FLUX.2-klein-base-9B); the [rank-64 character LoRA](https://huggingface.co/laion/FLUX.2-klein-base-9B-vivarium-loras) is linked with examples and inference instructions. Native generation uses **50 steps, real CFG 4 and BF16**.
+
+Named streaming configurations separate the original `base`, additional `humans_extra`, and `specials` families as their TARs become available. `default` reads all currently uploaded families. Demographic labels are intended prompt attributes; special creatures are not assigned human ethnicities. Use identity IDs for train/evaluation splitting.
+
+
 # Complete character collection and professional outfit plan
 
 **Planned total: 840 identities, 12,140 images.** The expression/special collection contains 7,500 planned images. A further **464 adult humans × 10 professional outfits = 4,640 sprites** reuse those identities and their neutral references; they do not add new personality IDs. See [all 100 jobs and the exact outfit allocation](JOB_OUTFITS.md). Job sprites uploaded so far: **16/4,640**. The `jobs` streaming configuration becomes available when its first TAR is uploaded.
@@ -325,7 +340,7 @@ dataset_info:
 The [teen/child detector recheck](SAFETY_RECHECK.md) reports actual image checksums, both LAION classifier scores, operational thresholds and any replacements. Intended ages, professions and demographic attributes describe prompting, not inferred real-world classifications.
 
 
-# Vivarium warm-painted greenscreen characters
+# Original 500-person collection
 
 **500 fictional people ×10 natural facial expressions =5,000 portrait images**, each832×1,248 pixels. The selected aesthetic is the **ghibli-warm-painted** direction: restrained facial proportions, delicate drawn lines, warm painted shading and a solid chroma-green backdrop. Generation uses the **undistilled9B base**, the **rank64 character LoRA**, **50 steps**, real **CFG4**, and **BF16**, with no quantization or distilled checkpoint.
 
@@ -432,7 +447,7 @@ image, metadata, prompt = sample["png"], sample["json"], sample["txt"]
 print(metadata["personality_id"], metadata["emotion"], image.size)
 ```
 
-The card config selects only `data/*.tar` so index files are not mistaken for training examples. An alternate standard-library TAR/HTTP reader avoids extracting image files:
+The `base` configuration selects only `data/*.tar`; the default also includes available extension TAR families. JSON indexes are excluded from training examples. An alternate standard-library TAR/HTTP reader avoids extracting image files:
 
 ```bash
 python tools/read_webdataset.py pilot-person-0253.tar \

metadata/job-outfits/progress.json

@@ -1,5 +1,5 @@
 {
-  "utc": "2026-10-06T23:51:37.450513+00:00",
+  "utc": "2026-10-07T00:04:17.530300+00:00",
   "jobs": {
     "done": 16,
     "queued": 4624
@@ -13,7 +13,7 @@
   "images_in_open_shards": 0,
   "images_currently_generating": 0,
   "inference_attempts_completed": 16,
-  "buffer_free_gb": 290.241298432,
+  "buffer_free_gb": 290.189635584,
   "images_uploaded": 16,
   "target_images": 4640
 }

metadata/minor-safety/public-stream-filter.json

@@ -1,6 +1,6 @@
 {
   "status": "restored_after_completed_recheck_and_all_replacement_uploads",
-  "utc": "2026-10-07T00:04:11.982813+00:00",
+  "utc": "2026-10-07T00:04:16.980601+00:00",
   "included_base_images": 5000,
   "withheld_base_images": 0
 }

tools/job_outfits.py

@@ -68,7 +68,8 @@ def worker(gpu,shared=False,limit=None):
     engine.event(c,gpu,'loaded');BUFFER.mkdir(parents=True,exist_ok=True)
     rows=list(c.execute('SELECT name FROM shards WHERE worker=?',(gpu,)))
     sequence=max([int(r['name'].removesuffix('.tar').rsplit('-',1)[1]) for r in rows]+[-1])+1
-    first=not rows;warm=False;completed=0;refs={}
+    first=not rows;warm=False;completed=0;refs={};scorer=None
+    from teen_generation_safety import in_scope,screen_frame,persist_actual_plan
     while limit is None or completed<limit:
         records=[];target=min(2 if first else 20,(limit-completed) if limit is not None else 20)
         tar=None
@@ -86,9 +87,20 @@ def worker(gpu,shared=False,limit=None):
             im=generate_native_image(pipe,prompt_embeds=embeds,seed=r['seed'],steps=50,guidance=4.,
                 negative_prompt=r['negative_prompt'],image=image,width=832,height=1248)[0]
             torch.cuda.synchronize();seconds=time.perf_counter()-t
+            safety=None
+            if in_scope(r):
+                if scorer is None:
+                    from laion_image_scorer import Scorer
+                    scorer=Scorer('cuda:0')
+                im,actual,safety,extra=screen_frame(pipe,r,r,im,image,scorer)
+                r.update({k:actual[k] for k in ['seed','negative_prompt']});seconds+=extra
+                persist_actual_plan(STATE,'image-plan.json','id',r)
+                c.execute('UPDATE jobs SET record=? WHERE id=?',(json.dumps(r,ensure_ascii=False,sort_keys=True),r['id']));c.commit()
             b=io.BytesIO();im.save(b,format='PNG',compress_level=4);png=b.getvalue()
             profile='shared-GPU-CPU-offload-job-pilot' if shared else 'dedicated-GPU-reference-job-outfits'
             m=metadata(r,dict(id=r['id'],seed=r['seed']),png,ref_sha,seconds,encode_seconds,name,profile);m.update(provenance)
+            if safety is not None:
+                safety['image_sha256']=m['image_sha256'];m['safety_check']=safety
             m['runtime']['includes_compile_warmup']=not warm;warm=True
             m['runtime']['gpu_shared_with_backgrounds']=False;m['runtime']['gpu_shared_with_character_production']=shared
             engine.add_member(tar,r['id']+'.png',png);engine.add_member(tar,r['id']+'.json',json.dumps(m,ensure_ascii=False,indent=2).encode());engine.add_member(tar,r['id']+'.txt',r['prompt'].encode())
@@ -100,6 +112,7 @@ def worker(gpu,shared=False,limit=None):
             for ident,seconds,m in records:c.execute("UPDATE jobs SET status='done',shard=?,generation_seconds=?,metadata=?,finished=? WHERE id=?",(name,seconds,m,engine.utc(),ident))
             c.commit();sequence+=1;first=False
         if not records:break
+    if scorer is not None:scorer.close()
     engine.event(c,gpu,'pilot_complete' if limit is not None else 'complete')
 
 if __name__=='__main__':