Spaces:
Paused
Paused
Daniel Marques
commited on
Commit
•
31c7d36
1
Parent(s):
302fc29
feat: add tranning in start container
Browse files- constants.py +5 -7
constants.py
CHANGED
@@ -37,15 +37,13 @@ MAX_NEW_TOKENS = CONTEXT_WINDOW_SIZE # int(CONTEXT_WINDOW_SIZE/4)
|
|
37 |
|
38 |
#### If you get a "not enough space in the buffer" error, you should reduce the values below, start with half of the original values and keep halving the value until the error stops appearing
|
39 |
|
40 |
-
N_GPU_LAYERS =
|
41 |
N_BATCH = CONTEXT_WINDOW_SIZE
|
42 |
|
43 |
### From experimenting with the Llama-2-7B-Chat-GGML model on 8GB VRAM, these values work:
|
44 |
# N_GPU_LAYERS = 20
|
45 |
# N_BATCH = 512
|
46 |
|
47 |
-
|
48 |
-
|
49 |
# https://python.langchain.com/en/latest/_modules/langchain/document_loaders/excel.html#UnstructuredExcelLoader
|
50 |
DOCUMENT_MAP = {
|
51 |
".txt": TextLoader,
|
@@ -100,11 +98,11 @@ EMBEDDING_MODEL_NAME = "hkunlp/instructor-large" # Uses 1.5 GB of VRAM (High Ac
|
|
100 |
#### (FOR GGUF MODELS)
|
101 |
####
|
102 |
|
103 |
-
|
104 |
-
|
105 |
|
106 |
-
MODEL_ID = "TheBloke/Llama-2-7b-Chat-GGUF"
|
107 |
-
MODEL_BASENAME = "llama-2-7b-chat.Q4_K_M.gguf"
|
108 |
|
109 |
# MODEL_ID = "TheBloke/Mistral-7B-Instruct-v0.1-GGUF"
|
110 |
# MODEL_BASENAME = "mistral-7b-instruct-v0.1.Q8_0.gguf"
|
|
|
37 |
|
38 |
#### If you get a "not enough space in the buffer" error, you should reduce the values below, start with half of the original values and keep halving the value until the error stops appearing
|
39 |
|
40 |
+
N_GPU_LAYERS = 100 # Llama-2-70B has 83 layers
|
41 |
N_BATCH = CONTEXT_WINDOW_SIZE
|
42 |
|
43 |
### From experimenting with the Llama-2-7B-Chat-GGML model on 8GB VRAM, these values work:
|
44 |
# N_GPU_LAYERS = 20
|
45 |
# N_BATCH = 512
|
46 |
|
|
|
|
|
47 |
# https://python.langchain.com/en/latest/_modules/langchain/document_loaders/excel.html#UnstructuredExcelLoader
|
48 |
DOCUMENT_MAP = {
|
49 |
".txt": TextLoader,
|
|
|
98 |
#### (FOR GGUF MODELS)
|
99 |
####
|
100 |
|
101 |
+
MODEL_ID = "TheBloke/Llama-2-13b-Chat-GGUF"
|
102 |
+
MODEL_BASENAME = "llama-2-13b-chat.Q4_K_M.gguf"
|
103 |
|
104 |
+
# MODEL_ID = "TheBloke/Llama-2-7b-Chat-GGUF"
|
105 |
+
# MODEL_BASENAME = "llama-2-7b-chat.Q4_K_M.gguf"
|
106 |
|
107 |
# MODEL_ID = "TheBloke/Mistral-7B-Instruct-v0.1-GGUF"
|
108 |
# MODEL_BASENAME = "mistral-7b-instruct-v0.1.Q8_0.gguf"
|