Instructions to use akatz-ai/MiniMax-H3-Character-Swap-LoRA with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusion Single File
How to use akatz-ai/MiniMax-H3-Character-Swap-LoRA with Diffusion Single File:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
h3 relay
Hi @whanajira , thanks for catching that issue so fast, forgot to include some local changes 😅
It should be available now on H3 Relay nodes main, commit: 924dfa94c280278abcd4bd62ebaae96fe4b10c7a
Please pull the latest changes and restart comfyui to load that node
Hi @whanajira , thanks for catching that issue so fast, forgot to include some local changes 😅
It should be available now on H3 Relay nodes main, commit: 924dfa94c280278abcd4bd62ebaae96fe4b10c7a
Please pull the latest changes and restart comfyui to load that node
sorry still not working tho.
gemini said :
Conclusion
To cut straight to the point, this is a programming bug caused by the latest updated h3-relay cache generation logic (removal_cache.py) attempting to use unhashable dict (dictionary) type data as a cache key.
Here is the detailed analysis of the cause:
Detailed Cause Analysis
Cache Key Generation Mechanism (ancestry function)
h3-relay tracks the structure (ancestry relationships) of models and input values to create a unique cache key so that identical computations can be reused for speed.
During this process, the ancestry function inside removal_cache.py recursively loops through the input values passed to the node and tries to bundle them all into a key (such as a frozenset).
Violation of Python's Hashing Rules (unhashable type: 'dict')
In Python, dictionaries (dict) are mutable objects whose values can be modified at any time, making them fundamentally ineligible for hashing (i.e., locking them into a fixed, unique value).
In the latest version of the code, a dictionary-type configuration dataset was included in a node's inputs. Because the code failed to filter this out and tried to bundle it into a hash key as-is, the Python interpreter raised a TypeError and halted.
Summary
You did not misconfigure anything. This is a source code defect caused by the h3-relay developer omitting error handling for dictionary-type inputs in the recent update.
Therefore, this issue cannot be resolved on your end. As guided earlier, you should revert to the previous stable commit (git checkout d732711) until the developer releases a patched update, which will allow it to function normally.
H3 Relay Windowed Edit · Character Swap still in red.
Hi @whanajira , just pushed another set of fixes to github, please try pulling the latest changes and restarting comfy. If there’s still issues I’ll need more info such as exact workflow and you comfyui version and system etc
Hi @whanajira , just pushed another set of fixes to github, please try pulling the latest changes and restarting comfy. If there’s still issues I’ll need more info such as exact workflow and you comfyui version and system etc
gg. no more errors. last pull fixed it. tyvm.
I still missing this nodes: H3RelayWindowedEdit, LabsH3PrepareVideo
It seems they are not in your node list:
H3 Relay · H3 Hybrid Model Loader
H3 Relay · H3 Model Loader
H3 Relay · FastH3 VSA Profile
H3 Relay · LTX Upscale Model Loader
H3 Relay · Pack LTX Model
H3 Relay · Cache Manager
H3 Relay · Apply Model LoRA
H3 Relay · Attention Backend
H3 Relay · Sequence Start
H3 Relay · Generate Shot
H3 Relay · Accepted Raw Latent
H3 Relay · H3 Ultimate 2× Enhance
H3 Relay · LTX 2× Enhance
H3 Relay · Interpolate
H3 Relay · Assemble Raw Sequence
H3 Relay · Assemble
Hi @Neuromaker , the code changes with those nodes are available on 'Nightly' in the manager or main branch of github. I've just bumped the version to 1.0.3 so the manager should have the latest nodes present once available.
Hi @Neuromaker , the code changes with those nodes are available on 'Nightly' in the manager or main branch of github. I've just bumped the version to 1.0.3 so the manager should have the latest nodes present once available.
Thx. Now it works!)
i wish i could have the same success but it fails for me in that spot over and over no matter what i try
why would this node Native loop · one H3 window throw error? no matter what I try on v2 workflow it’s the same error in every run. The numbers change, but the gap doesn’t.
Run Expected rows Got Missing
1344×736 (most runs) 37696 36730 966



