clag 0.0.4 → 0.0.5
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/README.md +9 -0
- data/lib/clag/version.rb +1 -1
- data/vendor/gems/sublayer/lib/sublayer/capabilities/llm_assistance.rb +50 -0
- metadata +1 -1
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA256:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: 7567d9817e4d7abc1c6da5ab57e36b99945fe64d3cb9d361d86c2c712e5e41af
|
|
4
|
+
data.tar.gz: 5425fc8a75d9267aaf341e4b13d82f76dc7dea19549fd2686d84bc667653f5f8
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: 8b629c28ad455cc20af1fe172c9dd774f88a980335c5f127fe7ebe6a624688f1ff004103261baedd252a7832beea01be378a2f1995e8c3c55cae7c1522355eb6
|
|
7
|
+
data.tar.gz: e2327b5d9f0625b0ddf0d084f40372e677d2952becd2ba7f0271ccbc8e234a0dd1e35a0231d2223c31ebdca881409c4d46cad8d72cc05b7dd23e1b0aea76e202
|
data/README.md
CHANGED
|
@@ -47,6 +47,15 @@ command with the help of an LLM!
|
|
|
47
47
|
|
|
48
48
|
* Select Groq as your preferred LLM by setting CLAG\_LLM=groq in your environment
|
|
49
49
|
|
|
50
|
+
### Using a Local Model
|
|
51
|
+
|
|
52
|
+
* Have a model locally from either Ollama or Llamafile with an OpenAI compatible
|
|
53
|
+
API
|
|
54
|
+
|
|
55
|
+
* Have the API server running on port 8080
|
|
56
|
+
|
|
57
|
+
* Select local as your preferred LLM by setting CLAG\_LLM=local in your environment
|
|
58
|
+
|
|
50
59
|
## Usage
|
|
51
60
|
|
|
52
61
|
Currently support one command: "g".
|
data/lib/clag/version.rb
CHANGED
|
@@ -25,6 +25,8 @@ module Sublayer
|
|
|
25
25
|
generate_with_claude
|
|
26
26
|
when "groq"
|
|
27
27
|
generate_with_groq
|
|
28
|
+
when "local"
|
|
29
|
+
generate_with_local_model
|
|
28
30
|
else
|
|
29
31
|
generate_with_openai
|
|
30
32
|
end
|
|
@@ -32,6 +34,54 @@ module Sublayer
|
|
|
32
34
|
|
|
33
35
|
private
|
|
34
36
|
|
|
37
|
+
def generate_with_local_model
|
|
38
|
+
system_prompt = <<-PROMPT
|
|
39
|
+
In this environment you have access to a set of tools you can use to answer the user's question.
|
|
40
|
+
|
|
41
|
+
You may call them like this:
|
|
42
|
+
<function_calls>
|
|
43
|
+
<invoke>
|
|
44
|
+
<tool_name>$TOOL_NAME</tool_name>
|
|
45
|
+
<parameters>
|
|
46
|
+
<command>value</command>
|
|
47
|
+
...
|
|
48
|
+
</parameters>
|
|
49
|
+
</invoke>
|
|
50
|
+
</function_calls>
|
|
51
|
+
|
|
52
|
+
Here are the tools available:
|
|
53
|
+
<tools>
|
|
54
|
+
#{self.class::OUTPUT_FUNCTION.to_xml}
|
|
55
|
+
</tools>
|
|
56
|
+
|
|
57
|
+
Respond only with valid xml.
|
|
58
|
+
The entire response should be wrapped in a <response> tag.
|
|
59
|
+
Any additional information not inside a tool call should go in a <scratch> tag.
|
|
60
|
+
PROMPT
|
|
61
|
+
|
|
62
|
+
response = HTTParty.post(
|
|
63
|
+
"http://localhost:8080/v1/chat/completions",
|
|
64
|
+
headers: {
|
|
65
|
+
"Authorization": "Bearer no-key",
|
|
66
|
+
"Content-Type": "application/json"
|
|
67
|
+
},
|
|
68
|
+
body: {
|
|
69
|
+
"model": "LLaMA_CPP",
|
|
70
|
+
"messages": [
|
|
71
|
+
{ "role": "system", "content": system_prompt },
|
|
72
|
+
{ "role": "user", "content": prompt }
|
|
73
|
+
]
|
|
74
|
+
}.to_json
|
|
75
|
+
)
|
|
76
|
+
|
|
77
|
+
text_containing_xml = JSON.parse(response.body).dig("choices", 0, "message", "content")
|
|
78
|
+
xml = text_containing_xml.match(/\<response\>(.*?)\<\/response\>/m).to_s
|
|
79
|
+
response_xml = Nokogiri::XML(xml)
|
|
80
|
+
function_output = response_xml.at_xpath("//parameters/command").children.to_s
|
|
81
|
+
|
|
82
|
+
return function_output
|
|
83
|
+
end
|
|
84
|
+
|
|
35
85
|
def generate_with_groq
|
|
36
86
|
system_prompt = <<-PROMPT
|
|
37
87
|
In this environment you have access to a set of tools you can use to answer the user's question.
|