Define four-profile production matrix with Medium default
This commit is contained in:
@@ -16,9 +16,10 @@ class Filter:
|
||||
class Valves(BaseModel):
|
||||
priority: int = 30
|
||||
fast_context_tokens: int = 76800
|
||||
medium_context_tokens: int = 94208
|
||||
long_context_tokens: int = 131072
|
||||
default_context_tokens: int = 76800
|
||||
medium_context_tokens: int = 160000
|
||||
large_context_tokens: int = 192000
|
||||
ultra_context_tokens: int = 262144
|
||||
default_context_tokens: int = 160000
|
||||
soft_context_ratio: float = 0.70
|
||||
hard_context_ratio: float = 0.84
|
||||
reserved_output_tokens: int = 8192
|
||||
@@ -107,8 +108,10 @@ class Filter:
|
||||
|
||||
def _context_limit(self, model: str) -> int:
|
||||
model = (model or "").lower()
|
||||
if "long" in model or "large" in model:
|
||||
return self.valves.long_context_tokens
|
||||
if "ultra" in model:
|
||||
return self.valves.ultra_context_tokens
|
||||
if "large" in model:
|
||||
return self.valves.large_context_tokens
|
||||
if "medium" in model:
|
||||
return self.valves.medium_context_tokens
|
||||
if "fast" in model:
|
||||
|
||||
Reference in New Issue
Block a user