Feedback and support: TensorBlock's Twitter/X, Telegram Group and Discord server
bigscience/bloomz-3b - GGUF
This repo contains GGUF format model files for bigscience/bloomz-3b.
The files were quantized using machines provided by TensorBlock, and they are compatible with llama.cpp as of commit b4011.
Prompt template
Model file specification
Filename | Quant type | File Size | Description |
---|---|---|---|
bloomz-3b-Q2_K.gguf | Q2_K | 1.516 GB | smallest, significant quality loss - not recommended for most purposes |
bloomz-3b-Q3_K_S.gguf | Q3_K_S | 1.707 GB | very small, high quality loss |
bloomz-3b-Q3_K_M.gguf | Q3_K_M | 1.905 GB | very small, high quality loss |
bloomz-3b-Q3_K_L.gguf | Q3_K_L | 2.016 GB | small, substantial quality loss |
bloomz-3b-Q4_0.gguf | Q4_0 | 2.079 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
bloomz-3b-Q4_K_S.gguf | Q4_K_S | 2.088 GB | small, greater quality loss |
bloomz-3b-Q4_K_M.gguf | Q4_K_M | 2.235 GB | medium, balanced quality - recommended |
bloomz-3b-Q5_0.gguf | Q5_0 | 2.428 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
bloomz-3b-Q5_K_S.gguf | Q5_K_S | 2.428 GB | large, low quality loss - recommended |
bloomz-3b-Q5_K_M.gguf | Q5_K_M | 2.546 GB | large, very low quality loss - recommended |
bloomz-3b-Q6_K.gguf | Q6_K | 2.799 GB | very large, extremely low quality loss |
bloomz-3b-Q8_0.gguf | Q8_0 | 3.621 GB | very large, extremely low quality loss - not recommended |
Downloading instruction
Command line
Firstly, install Huggingface Client
pip install -U "huggingface_hub[cli]"
Then, downoad the individual model file the a local directory
huggingface-cli download tensorblock/bloomz-3b-GGUF --include "bloomz-3b-Q2_K.gguf" --local-dir MY_LOCAL_DIR
If you wanna download multiple model files with a pattern (e.g., *Q4_K*gguf
), you can try:
huggingface-cli download tensorblock/bloomz-3b-GGUF --local-dir MY_LOCAL_DIR --local-dir-use-symlinks False --include='*Q4_K*gguf'
- Downloads last month
- 243
Model tree for tensorblock/bloomz-3b-GGUF
Base model
bigscience/bloomz-3bDataset used to train tensorblock/bloomz-3b-GGUF
Evaluation results
- Accuracy on Winogrande XL (xl)validation set self-reported53.670
- Accuracy on XWinograd (en)test set self-reported59.230
- Accuracy on XWinograd (fr)test set self-reported53.010
- Accuracy on XWinograd (jp)test set self-reported52.450
- Accuracy on XWinograd (pt)test set self-reported53.610
- Accuracy on XWinograd (ru)test set self-reported53.970
- Accuracy on XWinograd (zh)test set self-reported60.910
- Accuracy on ANLI (r1)validation set self-reported40.100
- Accuracy on ANLI (r2)validation set self-reported36.800
- Accuracy on ANLI (r3)validation set self-reported40.000