Loading
This repository holds the LoRA adapter; the full fine-tune of the same recipe is AbstractPhil/qwen3.5-0.8b-think-symbolic-image-full. Every Lucia loads the same way and is run the same way: thinking off, greedy decoding, no system prompt (the
row's form is learned, and a system turn she was not trained with makes her over-write). Needs transformers 5; the adapter also
needs peft 0.18 or newer.
The LoRA adapter (AbstractPhil/qwen3.5-0.8b-think-symbolic-image): the base model, the adapter merged into it, the
processor from the adapter repository (the base processor's files are kept there).
import torchfrom peft import PeftModelfrom transformers import AutoModelForMultimodalLM, AutoProcessor processor = AutoProcessor.from_pretrained("AbstractPhil/qwen3.5-0.8b-think-symbolic-image")model = AutoModelForMultimodalLM.from_pretrained("Qwen/Qwen3.5-0.8B", dtype=torch.bfloat16, device_map="cuda")model = PeftModel.from_pretrained(model, "AbstractPhil/qwen3.5-0.8b-think-symbolic-image").merge_and_unload().eval()
The full fine-tune (AbstractPhil/qwen3.5-0.8b-think-symbolic-image-full): the model and the processor from the full
repository, nothing else to merge.
import torchfrom transformers import AutoModelForMultimodalLM, AutoProcessor processor = AutoProcessor.from_pretrained("AbstractPhil/qwen3.5-0.8b-think-symbolic-image-full")model = AutoModelForMultimodalLM.from_pretrained("AbstractPhil/qwen3.5-0.8b-think-symbolic-image-full", dtype=torch.bfloat16, device_map="cuda").eval()
Then, for either:
from PIL import Image image = Image.open("picture.jpg").convert("RGB")prompt = ("Look at this picture and think it through in writing, a closer look, ending with a caption; then write a caption " "and a prompt or tags, and the questions a careful looker would still ask.")messages = [{"role": "user", "content": [{"type": "image", "image": image}, {"type": "text", "text": prompt}]}]text = processor.apply_chat_template(messages, add_generation_prompt=True, tokenize=False, enable_thinking=False)inputs = processor(text=[text], images=[image], return_tensors="pt").to(model.device)out = model.generate(**inputs, max_new_tokens=800, do_sample=False)print(processor.batch_decode(out[:, inputs["input_ids"].shape[1]:], skip_special_tokens=True)[0])
The depth words: "a quick look" (1 to 2 thoughts), "a closer look" (3 to 5), "a long look" (6 to 10). The parser, the scorer and
the trainer are in qwen-symbolic-thinking
(qwen_symbolic_thinking.chain.parse, chain.score); the same package extracted every number below.
The reads
Read the same way as Lucia v1, the 9B: 60 pictures and 40 specifications no adapter saw, against the dataset's own rows (row
shape, said / backed, definitions); then the medium term on two sets no training row came from: 500 National Gallery of Art
pictures (100 drawings, prints, paintings, photographs and objects each) and 200 pictures of AIC, NGA, Cleveland and the Met.
The 9B's numbers stand beside hers.
Table with columns: read, Lucia 0.8B, Lucia v1 (9B)| read | Lucia 0.8B | Lucia v1 (9B) |
|---|
| row shape, pictures / specifications | 1.00 / 1.00 | 1.00 / 1.00 |
| said / backed, whole rows | 0.19 / 0.30 | 0.19 / 0.30 |
| definitions, pictures / specifications | 0.00 / 0.12 | 0.02 / 0.08 |
| 500 held-out pictures: painting / print / drawing / photograph / object named rightly of 100 | 83 / 74 / 80 / 79 / 90 | 83 / 76 / 86 / 80 / 90 |
| 200 museum pictures: painting / print / drawing / photograph / object | 13 of 27 / 37 of 39 / 12 of 13 / 12 of 13 / 66 of 66 | 13 of 27 / 38 of 39 / 12 of 13 / 11 of 13 / 66 of 66 |
| the 200 museum pictures on one RTX PRO 6000, batch 64, 800 new tokens a row | 0.13 s a row | 0.27 s a row |
What this says: the smallest Lucia keeps the row's form whole, backs what she says at the 9B's level on the 60-row screen, and
names the medium a little less often (the drawing term 80 against 86, the print 74 against 76; the museum pictures at the 9B's
level). Where the size shows is the long specification rows, written from a text description alone: her thoughts there carry a
defining phrase in 50% of rows against the 9B's 35%, and her stock mood words come more often (47% of the 500 held-out rows
against 43%). She is a fair small looker, not the best one: the 4B and the 2B hold the medium term closer to the 9B.
A second arm tweaked for a small model (three epochs at lr 3e-4, the rest the same) was read beside her and not chosen: 82 / 73 /
79 / 80 / 91 on the 500 held-out pictures (405 against her 406), said / backed 0.20 / 0.28, and 3 of the 60 picture rows outside
their depth's range (0.95 against her 1.00). The verbatim recipe is the release; the tweaked arm stays in the training
repository. The museum painting row understates every size: watercolours and gouaches on paper are named drawings, as the
National Gallery's own catalogue does. The reads are made with qwen_symbolic_thinking.tools (probe_read, museum_read,
museum_mismatch, screen_conflicts), so any arm can be read the same way.
How she was trained
Table | |
|---|
| base | Qwen/Qwen3.5-0.8B, bf16, the vision tower frozen |
| adapter | LoRA rank 32, alpha 64, dropout 0.05 on the language model's q, k, v, o, gate, up, down, in_proj_qkv, in_proj_z and out_proj projections (20.4 M trainable parameters, saved in fp32) |
| data | Lucia v1's rows exactly: 951 picture rows taught twice, 500 long-look drawn rows, 2,000 synthetic closer looks and 17,385 museum rows whose catalogue gives the medium, the prompt line's first term set to it (6,606 drawings, 3,523 paintings, 3,000 prints, 2,016 photographs, 2,240 objects); 20,836 items, 21,787 examples; the 500 probe pictures held out |
| optimiser | plain Adam 2e-4, cosine schedule, 3% warmup, batch 8 x 2, two epochs (2,724 steps, 1.36 s a step, 62 minutes on one RTX PRO 6000), the loss on the answer only, seed 20261030 |
| thinking | off in training and at inference: the 0.8B answers without thinking by default, and every Lucia is taught and run with enable_thinking=False, so the smallest size learns the task's shape the same way as the largest |
| the arm not chosen | three epochs at lr 3e-4 (4,086 steps, 86 minutes), read beside this one and left in the training repository (the reads above) |
| record | qwen-symbolic-thinking v0.1.18, run_queue("lucia_0.8b") from notebooks/lucia_0.8b.ipynb; the run's full log is train_summary.json beside the weights |
Versions
Table with columns: version, where, tag| version | where | tag |
|---|
| Lucia 0.8B v1 | the root of this repository | lucia-v1 |
The family
The 9B is Lucia v1; every other size is trained under her recipe and read on the same held-out sets.
License
Apache-2.0, the base model's licence. The rows she was taught come from the datasets above under their own terms.