Spaces:
Running
on
T4
Running
on
T4
model changed to Haiku 3.5 for query re=write
Browse files
utilities/invoke_models.py
CHANGED
@@ -11,8 +11,6 @@ import streamlit as st
|
|
11 |
#from transformers import VisionEncoderDecoderModel, ViTImageProcessor, AutoTokenizer
|
12 |
#import torch
|
13 |
|
14 |
-
region = 'us-east-1'
|
15 |
-
|
16 |
def get_bedrock_client():
|
17 |
return boto3.client(
|
18 |
'bedrock-runtime',
|
@@ -32,8 +30,8 @@ def invoke_model(input):
|
|
32 |
}),
|
33 |
modelId="amazon.titan-embed-text-v1",
|
34 |
accept="application/json",
|
35 |
-
contentType="application/json"
|
36 |
-
performanceConfigLatency='optimized'
|
37 |
)
|
38 |
|
39 |
response_body = json.loads(response.get("body").read())
|
@@ -54,7 +52,7 @@ def invoke_model_mm(text,img):
|
|
54 |
contentType = "application/json"
|
55 |
|
56 |
response = bedrock_runtime_client.invoke_model(
|
57 |
-
body=body, modelId=modelId, accept=accept, contentType=contentType
|
58 |
)
|
59 |
response_body = json.loads(response.get("body").read())
|
60 |
#print(response_body)
|
@@ -63,7 +61,7 @@ def invoke_model_mm(text,img):
|
|
63 |
def invoke_llm_model(input,is_stream):
|
64 |
if(is_stream == False):
|
65 |
response = bedrock_runtime_client.invoke_model(
|
66 |
-
modelId= "anthropic.claude-3-5-haiku-20241022-v1:0",#"anthropic.claude-3-5-sonnet-20240620-v1:0",,
|
67 |
contentType = "application/json",
|
68 |
accept = "application/json",
|
69 |
performanceConfigLatency='optimized',
|
|
|
11 |
#from transformers import VisionEncoderDecoderModel, ViTImageProcessor, AutoTokenizer
|
12 |
#import torch
|
13 |
|
|
|
|
|
14 |
def get_bedrock_client():
|
15 |
return boto3.client(
|
16 |
'bedrock-runtime',
|
|
|
30 |
}),
|
31 |
modelId="amazon.titan-embed-text-v1",
|
32 |
accept="application/json",
|
33 |
+
contentType="application/json"
|
34 |
+
#,performanceConfigLatency='optimized'
|
35 |
)
|
36 |
|
37 |
response_body = json.loads(response.get("body").read())
|
|
|
52 |
contentType = "application/json"
|
53 |
|
54 |
response = bedrock_runtime_client.invoke_model(
|
55 |
+
body=body, modelId=modelId, accept=accept, contentType=contentType#,performanceConfigLatency='optimized'
|
56 |
)
|
57 |
response_body = json.loads(response.get("body").read())
|
58 |
#print(response_body)
|
|
|
61 |
def invoke_llm_model(input,is_stream):
|
62 |
if(is_stream == False):
|
63 |
response = bedrock_runtime_client.invoke_model(
|
64 |
+
modelId= "us.anthropic.claude-3-5-haiku-20241022-v1:0"#"anthropic.claude-3-5-haiku-20241022-v1:0",#"anthropic.claude-3-5-sonnet-20240620-v1:0",,
|
65 |
contentType = "application/json",
|
66 |
accept = "application/json",
|
67 |
performanceConfigLatency='optimized',
|