release/v1.4 -> master (#548)

This commit is contained in:
cybermaggedon 2025-10-06 17:54:26 +01:00 committed by GitHub
parent 3ec2cd54f9
commit 2bd68ed7f4
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
94 changed files with 8571 additions and 1740 deletions

View file

@ -32,7 +32,7 @@ class Processor(LlmService):
token = params.get("token", default_token)
temperature = params.get("temperature", default_temperature)
max_output = params.get("max_output", default_max_output)
model = default_model
model = params.get("model", default_model)
if endpoint is None:
raise RuntimeError("Azure endpoint not specified")
@ -53,9 +53,11 @@ class Processor(LlmService):
self.token = token
self.temperature = temperature
self.max_output = max_output
self.model = model
self.default_model = model
def build_prompt(self, system, content):
def build_prompt(self, system, content, temperature=None):
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
data = {
"messages": [
@ -67,7 +69,7 @@ class Processor(LlmService):
}
],
"max_tokens": self.max_output,
"temperature": self.temperature,
"temperature": effective_temperature,
"top_p": 1
}
@ -100,13 +102,22 @@ class Processor(LlmService):
return result
async def generate_content(self, system, prompt):
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
try:
prompt = self.build_prompt(
system,
prompt
prompt,
effective_temperature
)
response = self.call_llm(prompt)
@ -125,7 +136,7 @@ class Processor(LlmService):
text = resp,
in_token = inputtokens,
out_token = outputtokens,
model = self.model
model = model_name
)
return resp

View file

@ -54,7 +54,7 @@ class Processor(LlmService):
self.temperature = temperature
self.max_output = max_output
self.model = model
self.default_model = model
self.openai = AzureOpenAI(
api_key=token,
@ -62,14 +62,22 @@ class Processor(LlmService):
azure_endpoint = endpoint,
)
async def generate_content(self, system, prompt):
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
prompt = system + "\n\n" + prompt
try:
resp = self.openai.chat.completions.create(
model=self.model,
model=model_name,
messages=[
{
"role": "user",
@ -81,7 +89,7 @@ class Processor(LlmService):
]
}
],
temperature=self.temperature,
temperature=effective_temperature,
max_tokens=self.max_output,
top_p=1,
)
@ -97,7 +105,7 @@ class Processor(LlmService):
text = resp.choices[0].message.content,
in_token = inputtokens,
out_token = outputtokens,
model = self.model
model = model_name
)
return r

View file

@ -41,21 +41,29 @@ class Processor(LlmService):
}
)
self.model = model
self.default_model = model
self.claude = anthropic.Anthropic(api_key=api_key)
self.temperature = temperature
self.max_output = max_output
logger.info("Claude LLM service initialized")
async def generate_content(self, system, prompt):
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
try:
response = message = self.claude.messages.create(
model=self.model,
model=model_name,
max_tokens=self.max_output,
temperature=self.temperature,
temperature=effective_temperature,
system = system,
messages=[
{
@ -81,7 +89,7 @@ class Processor(LlmService):
text = resp,
in_token = inputtokens,
out_token = outputtokens,
model = self.model
model = model_name
)
return resp

View file

@ -39,21 +39,29 @@ class Processor(LlmService):
}
)
self.model = model
self.default_model = model
self.temperature = temperature
self.cohere = cohere.Client(api_key=api_key)
logger.info("Cohere LLM service initialized")
async def generate_content(self, system, prompt):
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
try:
output = self.cohere.chat(
model=self.model,
output = self.cohere.chat(
model=model_name,
message=prompt,
preamble = system,
temperature=self.temperature,
temperature=effective_temperature,
chat_history=[],
prompt_truncation='auto',
connectors=[]
@ -71,7 +79,7 @@ class Processor(LlmService):
text = resp,
in_token = inputtokens,
out_token = outputtokens,
model = self.model
model = model_name
)
return resp

View file

@ -53,10 +53,13 @@ class Processor(LlmService):
)
self.client = genai.Client(api_key=api_key)
self.model = model
self.default_model = model
self.temperature = temperature
self.max_output = max_output
# Cache for generation configs per model
self.generation_configs = {}
block_level = HarmBlockThreshold.BLOCK_ONLY_HIGH
self.safety_settings = [
@ -83,22 +86,45 @@ class Processor(LlmService):
logger.info("GoogleAIStudio LLM service initialized")
async def generate_content(self, system, prompt):
def _get_or_create_config(self, model_name, temperature=None):
"""Get or create generation config with dynamic temperature"""
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
generation_config = types.GenerateContentConfig(
temperature = self.temperature,
top_p = 1,
top_k = 40,
max_output_tokens = self.max_output,
response_mime_type = "text/plain",
system_instruction = system,
safety_settings = self.safety_settings,
)
# Create cache key that includes temperature to avoid conflicts
cache_key = f"{model_name}:{effective_temperature}"
if cache_key not in self.generation_configs:
logger.info(f"Creating generation config for '{model_name}' with temperature {effective_temperature}")
self.generation_configs[cache_key] = types.GenerateContentConfig(
temperature = effective_temperature,
top_p = 1,
top_k = 40,
max_output_tokens = self.max_output,
response_mime_type = "text/plain",
safety_settings = self.safety_settings,
)
return self.generation_configs[cache_key]
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
generation_config = self._get_or_create_config(model_name, effective_temperature)
# Set system instruction per request (can't be cached)
generation_config.system_instruction = system
try:
response = self.client.models.generate_content(
model=self.model,
model=model_name,
config=generation_config,
contents=prompt,
)
@ -114,7 +140,7 @@ class Processor(LlmService):
text = resp,
in_token = inputtokens,
out_token = outputtokens,
model = self.model
model = model_name
)
return resp

View file

@ -39,7 +39,7 @@ class Processor(LlmService):
}
)
self.model = model
self.default_model = model
self.llamafile=llamafile
self.temperature = temperature
self.max_output = max_output
@ -50,25 +50,33 @@ class Processor(LlmService):
logger.info("Llamafile LLM service initialized")
async def generate_content(self, system, prompt):
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
prompt = system + "\n\n" + prompt
try:
resp = self.openai.chat.completions.create(
model=self.model,
model=model_name,
messages=[
{"role": "user", "content": prompt}
]
#temperature=self.temperature,
#max_tokens=self.max_output,
#top_p=1,
#frequency_penalty=0,
#presence_penalty=0,
#response_format={
# "type": "text"
#}
],
temperature=effective_temperature,
max_tokens=self.max_output,
top_p=1,
frequency_penalty=0,
presence_penalty=0,
response_format={
"type": "text"
}
)
inputtokens = resp.usage.prompt_tokens
@ -82,7 +90,7 @@ class Processor(LlmService):
text = resp.choices[0].message.content,
in_token = inputtokens,
out_token = outputtokens,
model = "llama.cpp",
model = model_name,
)
return resp

View file

@ -39,7 +39,7 @@ class Processor(LlmService):
}
)
self.model = model
self.default_model = model
self.url = url + "v1/"
self.temperature = temperature
self.max_output = max_output
@ -50,7 +50,15 @@ class Processor(LlmService):
logger.info("LMStudio LLM service initialized")
async def generate_content(self, system, prompt):
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
prompt = system + "\n\n" + prompt
@ -59,18 +67,18 @@ class Processor(LlmService):
logger.debug(f"Prompt: {prompt}")
resp = self.openai.chat.completions.create(
model=self.model,
model=model_name,
messages=[
{"role": "user", "content": prompt}
]
#temperature=self.temperature,
#max_tokens=self.max_output,
#top_p=1,
#frequency_penalty=0,
#presence_penalty=0,
#response_format={
# "type": "text"
#}
],
temperature=effective_temperature,
max_tokens=self.max_output,
top_p=1,
frequency_penalty=0,
presence_penalty=0,
response_format={
"type": "text"
}
)
logger.debug(f"Full response: {resp}")
@ -86,7 +94,7 @@ class Processor(LlmService):
text = resp.choices[0].message.content,
in_token = inputtokens,
out_token = outputtokens,
model = self.model
model = model_name
)
return resp

View file

@ -41,21 +41,29 @@ class Processor(LlmService):
}
)
self.model = model
self.default_model = model
self.temperature = temperature
self.max_output = max_output
self.mistral = Mistral(api_key=api_key)
logger.info("Mistral LLM service initialized")
async def generate_content(self, system, prompt):
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
prompt = system + "\n\n" + prompt
try:
resp = self.mistral.chat.complete(
model=self.model,
model=model_name,
messages=[
{
"role": "user",
@ -67,7 +75,7 @@ class Processor(LlmService):
]
}
],
temperature=self.temperature,
temperature=effective_temperature,
max_tokens=self.max_output,
top_p=1,
frequency_penalty=0,
@ -87,7 +95,7 @@ class Processor(LlmService):
text = resp.choices[0].message.content,
in_token = inputtokens,
out_token = outputtokens,
model = self.model
model = model_name
)
return resp

View file

@ -17,6 +17,7 @@ from .... base import LlmService, LlmResult
default_ident = "text-completion"
default_model = 'gemma2:9b'
default_temperature = 0.0
default_ollama = os.getenv("OLLAMA_HOST", 'http://localhost:11434')
class Processor(LlmService):
@ -24,25 +25,36 @@ class Processor(LlmService):
def __init__(self, **params):
model = params.get("model", default_model)
temperature = params.get("temperature", default_temperature)
ollama = params.get("ollama", default_ollama)
super(Processor, self).__init__(
**params | {
"model": model,
"temperature": temperature,
"ollama": ollama,
}
)
self.model = model
self.default_model = model
self.temperature = temperature
self.llm = Client(host=ollama)
async def generate_content(self, system, prompt):
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
prompt = system + "\n\n" + prompt
try:
response = self.llm.generate(self.model, prompt)
response = self.llm.generate(model_name, prompt, options={'temperature': effective_temperature})
response_text = response['response']
logger.debug("Sending response...")
@ -55,7 +67,7 @@ class Processor(LlmService):
text = response_text,
in_token = inputtokens,
out_token = outputtokens,
model = self.model
model = model_name
)
return resp
@ -84,6 +96,13 @@ class Processor(LlmService):
help=f'ollama (default: {default_ollama})'
)
parser.add_argument(
'-t', '--temperature',
type=float,
default=default_temperature,
help=f'LLM temperature parameter (default: {default_temperature})'
)
def run():
Processor.launch(default_ident, __doc__)

View file

@ -47,7 +47,7 @@ class Processor(LlmService):
}
)
self.model = model
self.default_model = model
self.temperature = temperature
self.max_output = max_output
@ -58,14 +58,22 @@ class Processor(LlmService):
logger.info("OpenAI LLM service initialized")
async def generate_content(self, system, prompt):
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
prompt = system + "\n\n" + prompt
try:
resp = self.openai.chat.completions.create(
model=self.model,
model=model_name,
messages=[
{
"role": "user",
@ -77,7 +85,7 @@ class Processor(LlmService):
]
}
],
temperature=self.temperature,
temperature=effective_temperature,
max_tokens=self.max_output,
top_p=1,
frequency_penalty=0,
@ -97,7 +105,7 @@ class Processor(LlmService):
text = resp.choices[0].message.content,
in_token = inputtokens,
out_token = outputtokens,
model = self.model
model = model_name
)
return resp

View file

@ -30,32 +30,43 @@ class Processor(LlmService):
base_url = params.get("url", default_base_url)
temperature = params.get("temperature", default_temperature)
max_output = params.get("max_output", default_max_output)
model = params.get("model", "tgi")
super(Processor, self).__init__(
**params | {
"temperature": temperature,
"max_output": max_output,
"url": base_url,
"model": model,
}
)
self.base_url = base_url
self.temperature = temperature
self.max_output = max_output
self.default_model = model
self.session = aiohttp.ClientSession()
logger.info(f"Using TGI service at {base_url}")
logger.info("TGI LLM service initialized")
async def generate_content(self, system, prompt):
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
headers = {
"Content-Type": "application/json",
}
request = {
"model": "tgi",
"model": model_name,
"messages": [
{
"role": "system",
@ -67,7 +78,7 @@ class Processor(LlmService):
}
],
"max_tokens": self.max_output,
"temperature": self.temperature,
"temperature": effective_temperature,
}
try:
@ -96,7 +107,7 @@ class Processor(LlmService):
text = ans,
in_token = inputtokens,
out_token = outputtokens,
model = "tgi",
model = model_name,
)
return resp

View file

@ -45,24 +45,32 @@ class Processor(LlmService):
self.base_url = base_url
self.temperature = temperature
self.max_output = max_output
self.model = model
self.default_model = model
self.session = aiohttp.ClientSession()
logger.info(f"Using vLLM service at {base_url}")
logger.info("vLLM LLM service initialized")
async def generate_content(self, system, prompt):
async def generate_content(self, system, prompt, model=None, temperature=None):
# Use provided model or fall back to default
model_name = model or self.default_model
# Use provided temperature or fall back to default
effective_temperature = temperature if temperature is not None else self.temperature
logger.debug(f"Using model: {model_name}")
logger.debug(f"Using temperature: {effective_temperature}")
headers = {
"Content-Type": "application/json",
}
request = {
"model": self.model,
"model": model_name,
"prompt": system + "\n\n" + prompt,
"max_tokens": self.max_output,
"temperature": self.temperature,
"temperature": effective_temperature,
}
try:
@ -91,7 +99,7 @@ class Processor(LlmService):
text = ans,
in_token = inputtokens,
out_token = outputtokens,
model = self.model,
model = model_name,
)
return resp