ed
/
rwkv.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Update README.md

This commit is contained in:
saharNooby 2023-04-07 10:16:12 +04:00
parent e26b408ea7
commit edd57a186c
1 changed files with 13 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
README.md
View File

@ -10,7 +10,7 @@ This project provides [a C library rwkv.h](rwkv.h) and [a convinient Python wrap
**TODO (contributions welcome!)**: **TODO (contributions welcome!)**:
1. Measure latency and perplexity of different model sizes (169M to 14B) and data types (FP32, FP16, Q4_0, Q4_1) 1. Measure latency and perplexity of different model sizes (169M to 14B) and data types (FP32, FP16, Q4_0, Q4_1, Q4_1_O)
2. Test on Linux (including Colab) and MacOS 2. Test on Linux (including Colab) and MacOS
3. Make required memory calculation more robust (see #4) 3. Make required memory calculation more robust (see #4)
@ -68,7 +68,7 @@ If everything went OK, `librwkv.so` (Linux) or `rwkv.o` (MacOS) file should appe
```commandline ```commandline
# Windows # Windows
python rwkv\convert_rwkv_to_ggml.py C:\RWKV-4-Pile-169M-20220807-8023.pth C:\rwkv.cpp-169M.bin float16 python rwkv\convert_pytorch_to_ggml.py C:\RWKV-4-Pile-169M-20220807-8023.pth C:\rwkv.cpp-169M.bin float16
# Linux / MacOS # Linux / MacOS
python rwkv/convert_pytorch_to_ggml.py ~/Downloads/RWKV-4-Pile-169M-20220807-8023.pth ~/Downloads/rwkv.cpp-169M.bin float16 python rwkv/convert_pytorch_to_ggml.py ~/Downloads/RWKV-4-Pile-169M-20220807-8023.pth ~/Downloads/rwkv.cpp-169M.bin float16
@ -80,13 +80,17 @@ To convert the model into INT4 quantized format, run:
```commandline ```commandline
# Windows # Windows
python rwkv\quantize.py C:\rwkv.cpp-169M.bin C:\rwkv.cpp-169M-Q4_1.bin 3 python rwkv\quantize.py C:\rwkv.cpp-169M.bin C:\rwkv.cpp-169M-Q4_1_O.bin 4
# Linux / MacOS # Linux / MacOS
python rwkv/quantize.py ~/Downloads/rwkv.cpp-169M.bin ~/Downloads/rwkv.cpp-169M-Q4_1.bin 3 python rwkv/quantize.py ~/Downloads/rwkv.cpp-169M.bin ~/Downloads/rwkv.cpp-169M-Q4_1_O.bin 4
``` ```
Pass `2` for `Q4_0` format (smaller size, lower quality), `3` for `Q4_1` format (larger size, higher quality). Formats available:
- `4`: `Q4_1_O`, preserves outliers, best quality, very slow (as FP32).
- `3`: `Q4_1`, preserves range, poor quality, very fast (as FP16).
- `2`: `Q4_0`, worst quality, moderately fast (between FP16 and FP32).
### 4. Run the model ### 4. Run the model
@ -98,20 +102,20 @@ To generate some text, run:
```commandline ```commandline
# Windows # Windows
python rwkv\generate_completions.py C:\rwkv.cpp-169M-Q4_1.bin python rwkv\generate_completions.py C:\rwkv.cpp-169M-Q4_1_O.bin
# Linux / MacOS # Linux / MacOS
python rwkv/generate_completions.py ~/Downloads/rwkv.cpp-169M-Q4_1.bin python rwkv/generate_completions.py ~/Downloads/rwkv.cpp-169M-Q4_1_O.bin
``` ```
To chat with a bot, run: To chat with a bot, run:
```commandline ```commandline
# Windows # Windows
python rwkv\chat_with_bot.py C:\rwkv.cpp-169M-Q4_1.bin python rwkv\chat_with_bot.py C:\rwkv.cpp-169M-Q4_1_O.bin
# Linux / MacOS # Linux / MacOS
python rwkv/chat_with_bot.py ~/Downloads/rwkv.cpp-169M-Q4_1.bin python rwkv/chat_with_bot.py ~/Downloads/rwkv.cpp-169M-Q4_1_O.bin
``` ```
Edit [generate_completions.py](rwkv%2Fgenerate_completions.py) or [chat_with_bot.py](rwkv%2Fchat_with_bot.py) to change prompts and sampling settings. Edit [generate_completions.py](rwkv%2Fgenerate_completions.py) or [chat_with_bot.py](rwkv%2Fchat_with_bot.py) to change prompts and sampling settings.