90 lines
4.6 KiB
Markdown
90 lines
4.6 KiB
Markdown
# Two-Model Similarity Analysis
|
|
|
|
These nodes compare two safetensors models without merging them or writing an
|
|
output file. Separate nodes are provided for checkpoints, diffusion models,
|
|
text encoders, LoRAs, and embeddings.
|
|
|
|
`exclude_patterns` accepts one regex per line, or glob patterns when
|
|
`glob_patterns` is enabled. By default, matches are excluded from metrics and
|
|
topology counts. Enable `include_mode` to analyze only matches instead; an empty
|
|
include filter selects no tensors. The same field and matching syntax are used
|
|
in both modes.
|
|
|
|
## Outputs
|
|
|
|
`comparison_report` contains conventional numerical differences and topology
|
|
findings. Both reports use Markdown headings and tables. `cwb_report` separately contains diagnostics derived from the
|
|
similarity, alignment, and consensus stages of Consensus-Weighted Blending.
|
|
`documentation` contains this reference.
|
|
|
|
Two CSV-text outputs follow the existing three outputs, preserving their socket
|
|
positions: `layerwise_metrics_csv` and `layerwise_cwb_csv`. Connect them to a
|
|
text-saving node and use a `.csv` filename; analysis itself writes no files.
|
|
The CSV rows follow the same key order and displayed numeric precision as the
|
|
layerwise Markdown tables. Dtypes A/B, cosine mean/min/max, and A/B affinities
|
|
have separate columns. Undefined values remain `N/A`; an empty analysis produces
|
|
headers without data rows. Documentation-only mode returns empty CSV strings.
|
|
|
|
The reports begin with a global summary, followed by inferred blockwise values
|
|
and one entry for every comparable tensor. Block names are approximated from
|
|
the tensor-key hierarchy through its first numeric path component; they are not
|
|
an architecture-specific interpretation.
|
|
|
|
## Standard metrics
|
|
|
|
- MAE, MSE, and RMSE describe absolute error at the original weight scale.
|
|
- Maximum absolute difference identifies the largest individual change.
|
|
- Relative L2 is `||A-B|| / ((||A||+||B||)/2)` and is scale-aware.
|
|
- Cosine similarity compares direction; Pearson correlation compares centered
|
|
variation.
|
|
- Exact equality, sign agreement, norm ratio, and finite-value coverage expose
|
|
differences hidden by averages.
|
|
|
|
The global section reports both parameter-weighted metrics and equal-layer
|
|
averages. No arbitrary combined similarity score is assigned. Undefined
|
|
zero-norm or constant-tensor statistics are shown as `N/A`.
|
|
|
|
Only shared, same-shape floating tensors enter numerical aggregates. Missing
|
|
keys and shape mismatches are listed separately and are never padded, cropped,
|
|
or replaced with zeros. Non-floating tensors receive an exact-equality check.
|
|
Quantized models carrying ComfyUI quantization metadata are rejected because
|
|
encoded storage values do not represent directly comparable model weights.
|
|
|
|
## CWB diagnostics
|
|
|
|
The CWB report provides pairwise row-cosine statistics and each input's affinity
|
|
to both mean and median consensus. It stops before CWB's merge-weighting stage:
|
|
no thresholds, contribution weights, diversity weighting, vector
|
|
reconstitution, norm rescaling, or output writer are used.
|
|
|
|
Checkpoint, diffusion-model, and text-encoder tensors always retain their fixed
|
|
coordinate alignment. Embedding and LoRA nodes expose one optional
|
|
`cwb_similarity_alignment` control. When enabled, it performs
|
|
threshold-independent greedy one-to-one matching derived from CWB and reports
|
|
coverage, matched-score statistics, and improvement over index alignment. This
|
|
operation is quadratic and may be slow for large embeddings.
|
|
|
|
LoRA down/up tensors are treated as one logical layer. Rank-component matching
|
|
uses the product of down and up cosine similarity, including CWB's paired-sign
|
|
equivalence for comparison. No source tensor is modified.
|
|
|
|
## Memory and failure handling
|
|
|
|
Inputs always use Unified Efficient Loader's low-memory streaming mode. Matching
|
|
A/B tensors are loaded concurrently through bounded `async_stream()` pipelines
|
|
with `batch_size=1` and `prefetch_batches=1`, but no subsequent logical layer is
|
|
requested early. A complete LoRA down/up pair is loaded once and used for both
|
|
reports. CUDA inputs use pinned transfer while retaining the current CPU source
|
|
tensors for fallback.
|
|
|
|
Calculations use FP32. When CUDA processing runs out of memory, CUDA memory is
|
|
cleared and the current logical layer is retried once on CPU. After success or
|
|
failure, all tensor references for that layer are released and every loaded key
|
|
is marked processed exactly once before advancing. Both reports list every CPU
|
|
fallback. A failed CPU retry or any non-OOM error stops analysis with the layer
|
|
name in the error.
|
|
|
|
`top_weight_differences` bounds the scalar-coordinate list; every comparable
|
|
tensor is still present in the layerwise report. `DOCUMENTATION ONLY` loads no
|
|
model data.
|