r2s-cur-twemoji-67bab78-8bf9e764-47646604
completed —
Hypothesis
With the Twemoji icons that fit the codec added to the training split (held-out concepts excluded), and everything else as in v9, the render-to-program model transcribes the 339 validation renders with lower mean 72 px pixel error than v9 trained on the same machine and seed. Pass requires the paired 95% bootstrap interval on the per-icon reduction against that v9 rerun to exclude zero. Its greedy latency on the RTX 4080 is recorded.
Visual output
Written result
Hypothesis. With the Twemoji icons that fit the codec added to the training split (held-out concepts excluded), and everything else as in v9, the render-to-program model transcribes the 339 validation renders with lower mean 72 px pixel error than v9 trained on the same machine and seed. Pass requires the paired 95% bootstrap interval on the per-icon reduction against that v9 rerun to exclude zero. Its greedy latency on the RTX 4080 is recorded.
Result
| measure | value |
|---|---|
| icons evaluated | 339 (primary/validation) |
| model pixel error, mean [95% CI] | 0.0844 [0.0791, 0.0898] |
| model pixel error, median | 0.0801 |
| exact program rate | 0.012 |
| rendered rate | 1.000 |
| blank canvas pixel error | 0.1720 |
| nearest training icon pixel error | 0.0897 [0.0848, 0.0947] |
| error reduction vs nearest icon | 0.0054 [0.0009, 0.0098] |
| icons where model beats nearest icon | 195 |
| CLIP top-1, model greedy (32 Gate N icons) | 0.469 |
| CLIP top-1, nearest training icon | 0.344 |
| CLIP top-1, OmniSVG 4B zero-shot (Gate N) | 0.609 |
Resources
| measure | value |
|---|---|
| device | NVIDIA B300 SXM6 AC |
| parameters | 9,025,826 |
| train seconds | 5753 |
| peak VRAM GiB | 4.388385772705078 |
| latency ms/icon, batch 1, median | 4905.1 |
| latency ms/icon, batch 1, p95 | 5123.5 |
| throughput ms/icon, batch 64 | 124.1 |
| model calls per icon (decoding) | 921.0 |
| torch / CUDA | 2.14.1+cu130 / 13.0 |
Latency covers encoding and decoding to a validated token program; it excludes rasterising the SVG. Samples: samples.png, rows are reference, model, nearest training icon.
State transitions
- running2026-10-30T13:52:16Z
- completed2026-10-30T15:31:20Z
Run record
Verbatim from runs/r2s-cur-twemoji-67bab78-8bf9e764-47646604/run.yaml, the record committed before launch.
schema_version
2
run_id
r2s-cur-twemoji-67bab78-8bf9e764-47646604
state
completed
planned_at
2026-10-30T13:52:16Z
hypothesis
With the Twemoji icons that fit the codec added to the training split (held-out concepts excluded), and everything else as in v9, the render-to-program model transcribes the 339 validation renders with lower mean 72 px pixel error than v9 trained on the same machine and seed. Pass requires the paired 95% bootstrap interval on the per-icon reduction against that v9 rerun to exclude zero. Its greedy latency on the RTX 4080 is recorded.
parent_run
r2s-full-v9-colour
git_commit
67bab78fdc1a5143a1b131f59e45979d02cdf2c0
dirty_patch_sha256
6f6ecc7737307bb4…config
configs/render2svg/cur-twemoji.yaml
config_sha256
8bf9e7643e50e423…config_resolved
model
image_size
144
d_model
256
heads
8
encoder_layers
2
decoder_layers
6
feedforward
1024
dropout
0.1
metric
true
fourier
10
order
path
sources
0
training
steps
60000
batch_size
32
learning_rate
0.0005
weight_decay
0.01
warmup_steps
1000
seed
7001
eval_every
2000
eval_icons
64
train_icons
None
evaluate_on_train
false
augment_variants
0
augment_seed
9001
augment_mirror
0.5
augment_max_shift
48
augment_colour
0.1
augment_online
true
compose_probability
0.5
compose_parts
4
augment_original
0.1
loader_workers
16
extra_training
twemoji
bf16
true
final_eval_icons
339
clip_icons
32
dataset
pilot_config
configs/learning/openmoji-g1-geometric-gate-v16.yaml
cache_sha256
476466042da98d72…train_icons
4892
augmented_variants
0
extra_training
format
1
commit
b6b55fef1e8636b540a6d016a4729ca8cdf2e60b
size
144
palette
#000000, #ffffff, #92d3f5, #61b2e4, #ea5a47, #d22f27 … and 22 more
slots
128
excluded_bases
354
candidates
3375
source_files
4009
excluded_as_held_out
634
fit_and_kept
2211
cache
/home/eai-jdkxz/eai-team11/mojidiff/cache/render2svg-extra-twemoji-23878da09a9c.npz
cache_sha256
b263d60cbb1ed1ad…augmented_sha256
None
evaluated_on
primary/validation
seed
7001
determinism
seeded; cuDNN and SDPA kernels not forced deterministic
model_parameters
9025826
command
.venv/bin/python -m mojidiff.learning.render2svg --config configs/render2svg/cur-twemoji.yaml
outputs
run_dir
runs/r2s-cur-twemoji-67bab78-8bf9e764-47646604
checkpoints
/home/eai-jdkxz/eai-team11/mojidiff/cache/runs/r2s-cur-twemoji-67bab78-8bf9e764-47646604
notes
completed_at
2026-10-30T15:31:20Z
resource
device
NVIDIA B300 SXM6 AC
torch_version
2.14.1+cu130
cuda_version
13.0
python_version
3.12.3
peak_vram_gib
4.388385772705078
train_seconds
5753.471284666972
inference_ms_per_icon
4905.104981007753
inference_p95_ms_per_icon
5123.528375988826
icons_per_second
0.20386923498517054
batched_ms_per_icon
124.14704900038487
batched_icons_per_call
64
excludes
rendering the output SVG to pixels
baselines
blank_canvas_pixel_error
0.1720480552069557
nearest_training_icon_pixel_error
0.08973124639576185
gate_n_omnisvg_zero_shot_top1
0.609375
nearest_training_icon_clip_top1
0.34375
result
icons
339
model_pixel_error
0.08437308370170753, 0.07912901020487362, 0.08979159720406601
model_pixel_error_median
0.08013041317462921
exact_program_rate
0.011799410029498525
rendered_rate
1.0
blank_pixel_error
0.1720480552069557
selected_step
58000
nearest_training_icon_with_extra_pixel_error
0.0875778230639301, 0.08281926003772477, 0.09234717615560617
nearest_training_icon_pixel_error
0.08973124639576185, 0.0848205749792233, 0.09466393787711656
model_minus_baseline_error_reduction
0.0053581626940543345, 0.0008519524123823919, 0.009808966775227662
icons_model_beats_baseline
195
clip_retrieval
model_greedy
top1_rate
0.46875
top5_rate
0.90625
mean_rank
2.59375
clip_to_reference_mean
0.9070877749472857
chance_top1
0.03125
icons
32
nearest_training_icon
top1_rate
0.34375
top5_rate
0.59375
mean_rank
7.40625
clip_to_reference_mean
0.8630617223680019
chance_top1
0.03125
icons
32
gate_n_omnisvg_zero_shot_top1
0.609375
gate_n_omnisvg_best_of_12_top1
0.828125
model_calls_per_icon
921