Add new models, drop old models, remove default_max_tokens

Refs hex#4, hex#5

llm_perplexity.py

@@ -3,19 +3,17 @@
 from pydantic import Field, field_validator, model_validator
 from typing import Optional, List
 
+
 @llm.hookimpl
 def register_models(register):
     # https://docs.perplexity.ai/docs/model-cards
-    register(Perplexity("sonar-small-chat", default_max_tokens=16384), aliases=("pp-small-chat",))
-    register(Perplexity("sonar-small-online", default_max_tokens=12288), aliases=("pp-small-online",))
-    register(Perplexity("sonar-medium-chat", default_max_tokens=16384), aliases=("pp-medium-chat",))
-    register(Perplexity("sonar-medium-online", default_max_tokens=12288), aliases=("pp-medium-online",))
-    register(Perplexity("codellama-70b-instruct", default_max_tokens=16384), aliases=("pp-70b-instruct",))
-    register(Perplexity("mistral-7b-instruct", default_max_tokens=16384), aliases=("pp-7b-instruct",))
-    register(Perplexity("mixtral-8x7b-instruct", default_max_tokens=16384), aliases=("pp-8x7b-instruct",))
-    register(Perplexity("mixtral-8x22b-instruct", default_max_tokens=16384), aliases=("pp-8x22b-instruct",))
-    register(Perplexity("llama-3-8b-instruct", default_max_tokens=8192))
-    register(Perplexity("llama-3-70b-instruct", default_max_tokens=8192))
+    register(Perplexity("llama-3-sonar-small-32k-chat"))
+    register(Perplexity("llama-3-sonar-small-32k-online"))
+    register(Perplexity("llama-3-sonar-large-32k-chat"))
+    register(Perplexity("llama-3-sonar-large-32k-online"))
+    register(Perplexity("llama-3-8b-instruct"))
+    register(Perplexity("llama-3-70b-instruct"))
+    register(Perplexity("mixtral-8x7b-instruct"), aliases=("pp-8x7b-instruct",))
 
 
 class PerplexityOptions(llm.Options):
@@ -54,15 +52,6 @@ class PerplexityOptions(llm.Options):
         default=None,
     )
 
-    @field_validator("max_tokens")
-    def validate_max_tokens(cls, values):
-        max_tokens = values.get("max_tokens")
-        default_max_tokens = values.get("default_max_tokens")
-        if max_tokens is not None and default_max_tokens is not None:
-            if not (0 < max_tokens <= default_max_tokens):
-                raise ValueError(f"max_tokens must be in range 1-{default_max_tokens}")
-        return max_tokens
-
     @field_validator("temperature")
     @classmethod
     def validate_temperature(cls, temperature):
@@ -76,14 +65,14 @@ def validate_top_p(cls, top_p):
         if top_p is not None and not (0.0 <= top_p <= 1.0):
             raise ValueError("top_p must be in range 0.0-1.0")
         return top_p
-    
+
     @field_validator("top_k")
     @classmethod
     def validate_top_k(cls, top_k):
         if top_k is not None and top_k <= 0 or top_k > 2048:
             raise ValueError("top_k must be in range 0-2048")
         return top_k
-    
+
     @model_validator(mode="after")
     def validate_temperature_top_p(self):
         if self.temperature != 1.0 and self.top_p is not None:
@@ -97,12 +86,10 @@ class Perplexity(llm.Model):
     model_id = "perplexity"
     can_stream = True
 
-    class Options(PerplexityOptions):
-        ...
+    class Options(PerplexityOptions): ...
 
-    def __init__(self, model_id, default_max_tokens=None):
+    def __init__(self, model_id):
         self.model_id = model_id
-        self.default_max_tokens = default_max_tokens
 
     def build_messages(self, prompt, conversation) -> List[dict]:
         messages = []
@@ -145,8 +132,8 @@ def execute(self, prompt, stream, response, conversation):
                 for text in stream:
                     yield text.choices[0].delta.content
         else:
-           completion = client.chat.completions.create(**kwargs)
-           yield completion.choices[0].message.content
+            completion = client.chat.completions.create(**kwargs)
+            yield completion.choices[0].message.content
 
     def __str__(self):
         return f"Perplexity: {self.model_id}"