TheBloke commited on
Commit
9de56e8
1 Parent(s): be79a76

Initial GGML model commit

Browse files
Files changed (1) hide show
  1. README.md +54 -0
README.md ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ inference: false
3
+ license: other
4
+ ---
5
+
6
+ # Tim Dettmers' Guanaco 33B GGML
7
+
8
+ These files are GGML format model files for [Tim Dettmers' Guanaco 33B](https://huggingface.co/timdettmers/guanaco-33b-merged).
9
+
10
+ GGML files are for CPU inference using [llama.cpp](https://github.com/ggerganov/llama.cpp) and libraries and UIs which support this format, such as:
11
+ * [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
12
+ * [KoboldCpp](https://github.com/LostRuins/koboldcpp)
13
+ * [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)
14
+ * [ctransformers](https://github.com/marella/ctransformers)
15
+
16
+ ## Other repositories available
17
+
18
+ * [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/guanaco-33B-GPTQ)
19
+ * [4-bit, 5-bit, and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/guanaco-33B-GGML)
20
+ * [Original unquantised fp16 model in HF format](https://huggingface.co/timdettmers/guanaco-33b-merged)
21
+
22
+ ## THE FILES IN MAIN BRANCH REQUIRES LATEST LLAMA.CPP (May 19th 2023 - commit 2d5db48)!
23
+
24
+ llama.cpp recently made another breaking change to its quantisation methods - https://github.com/ggerganov/llama.cpp/pull/1508
25
+
26
+ I have quantised the GGML files in this repo with the latest version. Therefore you will require llama.cpp compiled on May 19th or later (commit `2d5db48` or later) to use them.
27
+
28
+ ## Provided files
29
+ | Name | Quant method | Bits | Size | RAM required | Use case |
30
+ | ---- | ---- | ---- | ---- | ---- | ----- |
31
+ | guanaco-33B.ggmlv3.q4_0.bin | q4_0 | 4 | 18.30 GB | 20.80 GB | 4-bit. |
32
+
33
+
34
+ ## How to run in `llama.cpp`
35
+
36
+ I use the following command line; adjust for your tastes and needs:
37
+
38
+ ```
39
+ ./main -t 12 -m guanaco-33B.v3.q5_0.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "Below is an instruction that describes a task. Write a response that appropriately completes the request.
40
+ ### Instruction:
41
+ Write a story about llamas
42
+ ### Response:"
43
+ ```
44
+ Change `-t 12` to the number of physical CPU cores you have. For example if your system has 8 cores/16 threads, use `-t 8`.
45
+
46
+ If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
47
+
48
+ ## How to run in `text-generation-webui`
49
+
50
+ Further instructions here: [text-generation-webui/docs/llama.cpp-models.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp-models.md).
51
+
52
+ Note: at this time text-generation-webui may not support the new May 19th llama.cpp quantisation methods for q4_0, q4_1 and q8_0 files.
53
+
54
+ # Original model card: Tim Dettmers' Guanaco 33B