Spaces:
Paused
Paused
Update constants.py
Browse files- constants.py +6 -6
constants.py
CHANGED
@@ -61,13 +61,13 @@ DOCUMENT_MAP = {
|
|
61 |
}
|
62 |
|
63 |
# Default Instructor Model
|
64 |
-
EMBEDDING_MODEL_NAME = "hkunlp/instructor-large" # Uses 1.5 GB of VRAM (High Accuracy with lower VRAM usage)
|
65 |
|
66 |
####
|
67 |
#### OTHER EMBEDDING MODEL OPTIONS
|
68 |
####
|
69 |
|
70 |
-
|
71 |
# EMBEDDING_MODEL_NAME = "intfloat/e5-large-v2" # Uses 1.5 GB of VRAM (A little less accurate than instructor-large)
|
72 |
# EMBEDDING_MODEL_NAME = "intfloat/e5-base-v2" # Uses 0.5 GB of VRAM (A good model for lower VRAM GPUs)
|
73 |
# EMBEDDING_MODEL_NAME = "all-MiniLM-L6-v2" # Uses 0.2 GB of VRAM (Less accurate but fastest - only requires 150mb of vram)
|
@@ -100,11 +100,11 @@ EMBEDDING_MODEL_NAME = "hkunlp/instructor-large" # Uses 1.5 GB of VRAM (High Ac
|
|
100 |
#### (FOR GGUF MODELS)
|
101 |
####
|
102 |
|
103 |
-
MODEL_ID = "TheBloke/Llama-2-13b-Chat-GGUF"
|
104 |
-
MODEL_BASENAME = "llama-2-13b-chat.Q4_K_M.gguf"
|
105 |
|
106 |
-
|
107 |
-
|
108 |
|
109 |
# MODEL_ID = "TheBloke/Mistral-7B-Instruct-v0.1-GGUF"
|
110 |
# MODEL_BASENAME = "mistral-7b-instruct-v0.1.Q8_0.gguf"
|
|
|
61 |
}
|
62 |
|
63 |
# Default Instructor Model
|
64 |
+
# EMBEDDING_MODEL_NAME = "hkunlp/instructor-large" # Uses 1.5 GB of VRAM (High Accuracy with lower VRAM usage)
|
65 |
|
66 |
####
|
67 |
#### OTHER EMBEDDING MODEL OPTIONS
|
68 |
####
|
69 |
|
70 |
+
EMBEDDING_MODEL_NAME = "hkunlp/instructor-xl" # Uses 5 GB of VRAM (Most Accurate of all models)
|
71 |
# EMBEDDING_MODEL_NAME = "intfloat/e5-large-v2" # Uses 1.5 GB of VRAM (A little less accurate than instructor-large)
|
72 |
# EMBEDDING_MODEL_NAME = "intfloat/e5-base-v2" # Uses 0.5 GB of VRAM (A good model for lower VRAM GPUs)
|
73 |
# EMBEDDING_MODEL_NAME = "all-MiniLM-L6-v2" # Uses 0.2 GB of VRAM (Less accurate but fastest - only requires 150mb of vram)
|
|
|
100 |
#### (FOR GGUF MODELS)
|
101 |
####
|
102 |
|
103 |
+
# MODEL_ID = "TheBloke/Llama-2-13b-Chat-GGUF"
|
104 |
+
# MODEL_BASENAME = "llama-2-13b-chat.Q4_K_M.gguf"
|
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"
|