Load Anima LoRA on 3.8B:
Anima38LoRALoader lets you use an Anima Base or 2.9B LoRA with an Anima 3.8B model. It remaps the LoRA’s model-layer weights to the matching layers inherited by the 3.8B model, rather than treating the two model sizes as interchangeable. It can also apply any text-encoder LoRA weights to the accompanying CLIP component. This gives you a way to bring compatible LoRA effects into an Anima 3.8B workflow while controlling their influence on the image model and text encoder separately.
Load Anima LoRA on 3.8B Input Parameters:
model
The MODEL you want to modify with the LoRA. This must be an Anima 3.8B model whose configuration reports 52 blocks; the node checks that block count before loading the LoRA and raises an error if it does not match. Connect the model you intend to use for image generation. No numeric range or default is defined for this input.
lora_name
Selects the LoRA file to load from the available files in ComfyUI’s loras locations. The node loads the selected file and remaps its applicable model-layer weights for the 3.8B model; it also passes along any applicable text-encoder weights. The options are the LoRA filenames available in your installation, so there is no fixed list or default specified in the source.
strength_model
Controls how strongly the remapped LoRA affects the image model. The default is 1.0; the allowed range is -100.0 to 100.0, in steps of 0.01. Set it to 0 to disable the LoRA’s model-side effect, or adjust it to change the strength passed to the LoRA loader. The node does not define a recommended value beyond its default.
clip
The CLIP component to which applicable text-encoder LoRA weights can be applied. Connect the CLIP component used alongside your model so the node can return the corresponding updated component for downstream prompt encoding. If the selected LoRA has no applicable text-encoder weights, there may be no text-encoder change to apply. No numeric range or default is defined for this input.
strength_clip
Controls how strongly applicable text-encoder LoRA weights affect CLIP, independently of strength_model. The default is 1.0; the allowed range is -100.0 to 100.0, in steps of 0.01. Set it to 0 when you do not want a CLIP-side LoRA effect. If both strength settings are 0, the node returns the incoming model and CLIP without checking the model’s block count or loading the LoRA file.
Load Anima LoRA on 3.8B Output Parameters:
model
The MODEL to use downstream after the selected LoRA’s remapped model-layer weights have been applied at strength_model. This output carries the model-side result for subsequent generation steps. When both strength settings are 0, it is the incoming model returned unchanged.
clip
The CLIP to use downstream after any applicable text-encoder LoRA weights have been applied at strength_clip. Connect this output to downstream prompt-encoding nodes to use the CLIP-side result. When both strength settings are 0, it is the incoming CLIP returned unchanged.
Load Anima LoRA on 3.8B Usage Tips:
- Connect an Anima 3.8B model configured with 52 blocks. The node checks this requirement before loading a LoRA whenever an effect is enabled.
- Choose an Anima Base or 2.9B LoRA from the
lora_namemenu. Keep the returnedmodelandclipconnected to their respective downstream nodes so you use the results of this loader. - Start with the default strengths of
1.0and adjust the model and CLIP strengths separately. Setstrength_clipto0if you only want a model-side effect, orstrength_modelto0if you only want to apply available text-encoder weights. - Set both strengths to
0when you want to bypass LoRA application temporarily. In that case, the node returns its incoming components without reading the selected LoRA file.
Load Anima LoRA on 3.8B Common Errors and Solutions:
Anima 3.8B LoRA Bridge requires a 52-block MODEL, got …
- Explanation: The connected model’s reported block count is not
52. The node cannot proceed with LoRA loading on that model. - Solution: Connect the intended Anima 3.8B model and check that you have not accidentally supplied a different model. If you only want to bypass this node, set both strengths to
0.
LoRA file not found or path resolution error
- Explanation: The selected
lora_namecannot be resolved to a file in the configuredloraslocations. The exact error wording depends on the file-path system. - Solution: Confirm that the LoRA file is in a configured LoRA folder, refresh the file list if needed, and select its current name from the menu.
LoRA loading or file-format error
- Explanation: The node could not read the selected LoRA file, so it cannot remap or apply its weights. The exact message depends on the file loader and the file’s contents.
- Solution: Check that the file is a valid, readable LoRA file. If it may be incomplete or damaged, replace it with a known-good copy and select it again.
LoRA remapping or application error
- Explanation: The selected file could not be remapped or applied successfully. The exact message depends on the LoRA’s contents and the underlying remapping or loading code; a filename appearing in the menu does not establish compatibility.
- Solution: Check that you selected an Anima Base or 2.9B LoRA intended for this bridge, and try a known-compatible LoRA to narrow down whether the file is the cause.
