matitos/matitos_news
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Temperature and seed LLM

Browse Source
This commit is contained in:
Luciano Gervasoni
2025-04-23 17:46:47 +02:00
parent cf55c586f7
commit b8fdcae5ec
1 changed files with 3 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
app_urls/fetcher/src/llm.py
View File

@@ -9,6 +9,7 @@ class OllamaClient():
def __init__(self): def __init__(self):
self.host = os.getenv("ENDPOINT_OLLAMA", "https://ollamamodel.matitos.org") self.host = os.getenv("ENDPOINT_OLLAMA", "https://ollamamodel.matitos.org")
self.client = ollama.Client(host=self.host) self.client = ollama.Client(host=self.host)
self.options = {"temperature": 0, "seed": 13579}
def _get_default_model(self): def _get_default_model(self):
return os.getenv("OLLAMA_MODEL_DEFAULT", "llama3.2:3b") return os.getenv("OLLAMA_MODEL_DEFAULT", "llama3.2:3b")
@@ -44,7 +45,7 @@ class OllamaClient():
def generate(self, model, prompt, format=None): def generate(self, model, prompt, format=None):
try: try:
# Generate response # Generate response
response = self.client.generate(model=model, prompt=prompt, format=format) response = self.client.generate(model=model, prompt=prompt, format=format, options=self.options)
# Extract response # Extract response
response = response.response response = response.response
# Json? -> Dict # Json? -> Dict
@@ -65,7 +66,7 @@ class OllamaClient():
def generate_stream(self, model, prompt): def generate_stream(self, model, prompt):
try: try:
# Generate response # Generate response
response = self.client.generate(model=model, prompt=prompt, format="json", stream=True) response = self.client.generate(model=model, prompt=prompt, format="json", stream=True, options=self.options)
# Streamed chunks # Streamed chunks
for chunk in response: for chunk in response:
yield chunk.response yield chunk.response