ImAXISOG commited on
Commit
524ebd9
·
verified ·
1 Parent(s): 3f8a561

Model card: which model to use on Mac and Windows, by memory

Browse files
Files changed (1) hide show
  1. README.md +17 -0
README.md CHANGED
@@ -108,6 +108,23 @@ It reads English, Hindi (Devanagari) and Hinglish (Hindi typed in English letter
108
  | **Based on** | Google's [google/gemma-4-E2B-it](https://huggingface.co/google/gemma-4-E2B-it), fine-tuned with LoRA |
109
  | **License** | Apache 2.0, a modified version of Gemma 4 (see [License](#license)) |
110
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
111
  ## Results
112
 
113
  ![How often the answer was good enough: untrained Gemma 4 E4B-it, Iris Lighter and Iris Best quality on six jobs](https://huggingface.co/ImAXISOG/iris-gemma-4-e2b-it-mlx/resolve/main/assets/quality.png)
 
108
  | **Based on** | Google's [google/gemma-4-E2B-it](https://huggingface.co/google/gemma-4-E2B-it), fine-tuned with LoRA |
109
  | **License** | Apache 2.0, a modified version of Gemma 4 (see [License](#license)) |
110
 
111
+ ## Which one should I use?
112
+
113
+ | Your computer | Use | Memory it needs |
114
+ |---|---|---|
115
+ | Mac with 16 GB or more | Best quality: [iris-gemma-4-e4b-it-mlx](https://huggingface.co/ImAXISOG/iris-gemma-4-e4b-it-mlx) | 3.6 GB at peak |
116
+ | Mac with 8 GB | Lighter: [iris-gemma-4-e2b-it-mlx](https://huggingface.co/ImAXISOG/iris-gemma-4-e2b-it-mlx) | 2.9 GB at peak |
117
+ | Windows PC with 16 GB or more | Best quality: [iris-gemma-4-e4b-it-gguf](https://huggingface.co/ImAXISOG/iris-gemma-4-e4b-it-gguf) | 3.9 GB, plus the 4.6 GB file kept in memory: 8.4 GB in all |
118
+ | Windows PC with 8 GB | Lighter: [iris-gemma-4-e2b-it-gguf](https://huggingface.co/ImAXISOG/iris-gemma-4-e2b-it-gguf) | 1.9 GB, plus the 2.8 GB file kept in memory: 4.8 GB in all |
119
+
120
+ If Hindi or Hinglish matters most, use Best quality whenever the computer has 16 GB: on Hindi Fix it is accepted
121
+ 95% of the time against Lighter's 71%. On Windows the model file counts too: when memory runs short, Windows drops
122
+ the file from memory and reads it from disk again, and answers slow down. That is why Best quality needs a 16 GB PC
123
+ even though its own memory is under 4 GB.
124
+
125
+ The Mac rows are the rule Iris uses at setup. The Windows numbers were measured with llama.cpp on a Mac's processor
126
+ cores. Neither has been confirmed yet on a real Windows PC or on an 8 GB computer of either kind.
127
+
128
  ## Results
129
 
130
  ![How often the answer was good enough: untrained Gemma 4 E4B-it, Iris Lighter and Iris Best quality on six jobs](https://huggingface.co/ImAXISOG/iris-gemma-4-e2b-it-mlx/resolve/main/assets/quality.png)