Pomni commited on
Commit
0aa5176
·
verified ·
1 Parent(s): 54e63ff

Update quant list

Browse files
Files changed (1) hide show
  1. README.md +5 -1
README.md CHANGED
@@ -27,7 +27,7 @@ This is a repository of **GGML quants for [distil-medium.en](https://huggingface
27
 
28
  If you are looking for a program to run this model with, then I would recommend [EasyWhisper UI](https://github.com/mehtabmahir/easy-whisper-ui), as it is user-friendly, has a GUI, and automates a lot of the hard stuff for you.
29
  ## List of Quants
30
- Sorry, but I am still in the middle of creating quants right now. For now, you can download the official F32 and F16 quants down below.
31
 
32
  Clicking on a link will download the corresponding quant instantly.
33
  | Link | Quant | Size | Notes
@@ -35,7 +35,11 @@ Clicking on a link will download the corresponding quant instantly.
35
  | [GGML](https://huggingface.co/distil-whisper/distil-medium.en/resolve/main/ggml-medium-32-2.en.fp32.bin) | F32 | 1.58 GB | Likely overkill. |
36
  | [GGML](https://huggingface.co/distil-whisper/distil-medium.en/resolve/main/ggml-medium-32-2.en.bin) | F16 | 794 MB | Performs better than Q8_0 for noisy audio and music. |
37
  | [GGML](https://huggingface.co/Pomni/distil-medium.en-ggml-allquants/resolve/main/ggml-distil-medium.en-q8_0.bin) | Q8_0 | 430 MB | Sweet spot; superficial quality loss at nearly double the speed. |
 
 
 
38
  | [GGML](https://huggingface.co/Pomni/distil-medium.en-ggml-allquants/resolve/main/ggml-distil-medium.en-q5_0.bin) | Q5_0 | 284 MB | Last "good" quant; anything below loses quality rapidly. |
 
39
 
40
  The F32 quant was taken from [distil-whisper/distil-medium.en/ggml-medium-32-2.en.fp32.bin](https://huggingface.co/distil-whisper/distil-medium.en/blob/main/ggml-medium-32-2.en.fp32.bin), and the F16 quant was taken from [distil-whisper/distil-medium.en/ggml-medium-32-2.en.bin](https://huggingface.co/distil-whisper/distil-medium.en/blob/main/ggml-medium-32-2.en.bin).
41
  ## Questions you may have
 
27
 
28
  If you are looking for a program to run this model with, then I would recommend [EasyWhisper UI](https://github.com/mehtabmahir/easy-whisper-ui), as it is user-friendly, has a GUI, and automates a lot of the hard stuff for you.
29
  ## List of Quants
30
+ I am still in the middle of creating quants right now. The most important quants that I deem to be "useful in real-world usage" (being able to provide acceptably accurate transcriptions) are already up however.
31
 
32
  Clicking on a link will download the corresponding quant instantly.
33
  | Link | Quant | Size | Notes
 
35
  | [GGML](https://huggingface.co/distil-whisper/distil-medium.en/resolve/main/ggml-medium-32-2.en.fp32.bin) | F32 | 1.58 GB | Likely overkill. |
36
  | [GGML](https://huggingface.co/distil-whisper/distil-medium.en/resolve/main/ggml-medium-32-2.en.bin) | F16 | 794 MB | Performs better than Q8_0 for noisy audio and music. |
37
  | [GGML](https://huggingface.co/Pomni/distil-medium.en-ggml-allquants/resolve/main/ggml-distil-medium.en-q8_0.bin) | Q8_0 | 430 MB | Sweet spot; superficial quality loss at nearly double the speed. |
38
+ | [GGML](https://huggingface.co/Pomni/distil-medium.en-ggml-allquants/resolve/main/ggml-distil-medium.en-q6_k.bin) | Q6_K | 336 MB | |
39
+ | [GGML](https://huggingface.co/Pomni/distil-medium.en-ggml-allquants/resolve/main/ggml-distil-medium.en-q5_k.bin) | Q5_K | 284 MB | |
40
+ | [GGML](https://huggingface.co/Pomni/distil-medium.en-ggml-allquants/resolve/main/ggml-distil-medium.en-q5_1.bin) | Q5_1 | 308 MB | |
41
  | [GGML](https://huggingface.co/Pomni/distil-medium.en-ggml-allquants/resolve/main/ggml-distil-medium.en-q5_0.bin) | Q5_0 | 284 MB | Last "good" quant; anything below loses quality rapidly. |
42
+ | [GGML](https://huggingface.co/Pomni/distil-medium.en-ggml-allquants/resolve/main/ggml-distil-medium.en-q4_k.bin) | Q4_K | 235 MB | *Might* not have lost too much quality, but I'm not sure. |
43
 
44
  The F32 quant was taken from [distil-whisper/distil-medium.en/ggml-medium-32-2.en.fp32.bin](https://huggingface.co/distil-whisper/distil-medium.en/blob/main/ggml-medium-32-2.en.fp32.bin), and the F16 quant was taken from [distil-whisper/distil-medium.en/ggml-medium-32-2.en.bin](https://huggingface.co/distil-whisper/distil-medium.en/blob/main/ggml-medium-32-2.en.bin).
45
  ## Questions you may have