Darwin uniform-soup of 3 Spark2.5-1.7B donors (226/226 tensors verified). Agentic/tool-use + uncensored + multilingual blend. Distillation donor for the 4B program. Needs transformers==5.17.0 + trustremotecode=True.
Independent publisher
Charlie
MC7ever
new emerging techniques, converting to mamba3, bootleg operations, fine tunes, inference and training all on shitty hardware
Models in Library1
Datasets in Library1
Models on Hugging Face5
Followers31
Models
Datasets
normalized and stored as per-dataset Parquet files under data/. video (bytes, capped) and tabular/timeseries (serialized to text). https://github.com/shadyuwugurl/hf-datasets-trees/blob/main/datasets.txt) - streaming=True reads via datasets.loaddataset; split fallback (train -> test -> validation -> first available) and per-row format auto-detect (dpo / video / vision / speech / chatml / sharegpt / alpaca / rawtext) - English filter (ASCII >= 60%) for text-only rows; media rows are always kept - Text truncated at 30k chars; images capped at 10MB (PNG), audio at 20MB (WAV), video at 25MB; 8 workers / 5 concurrent streams, background push every 6 min - Resume matches filenames already in…