Tree: Add generation logging support

Generations can be logged in the console along with sampling parameters
if the user enables it in config.

Metrics are always logged at the end of each prompt. In addition,
the model endpoint tells the user if they're being logged or not
for transparancy purposes.

Signed-off-by: kingbri <bdashore3@proton.me>

OAI/types/model.py

@@ -1,6 +1,7 @@
 from pydantic import BaseModel, Field
 from time import time
 from typing import List, Optional
+from gen_logging import LogConfig
 
 class ModelCardParameters(BaseModel):
     max_seq_len: Optional[int] = 4096
@@ -14,6 +15,7 @@ class ModelCard(BaseModel):
     object: str = "model"
     created: int = Field(default_factory=lambda: int(time()))
     owned_by: str = "tabbyAPI"
+    logging: Optional[LogConfig] = None
     parameters: Optional[ModelCardParameters] = None
 
 class ModelList(BaseModel):

config_sample.yml

@@ -13,6 +13,14 @@ network:
   # The port to host on (default: 5000)
   port: 5000
 
+# Options for logging
+logging:
+  # Enable prompt logging (default: False)
+  prompt: False
+
+  # Enable generation parameter logging (default: False)
+  generation_params: False
+
 # Options for model overrides and loading
 model:
   # Overrides the directory to look for models (default: models)

gen_logging.py

@@ -0,0 +1,47 @@
+from typing import Dict
+from pydantic import BaseModel
+
+# Logging preference config
+class LogConfig(BaseModel):
+    prompt: bool = False
+    generation_params: bool = False
+
+# Global reference to logging preferences
+config = LogConfig()
+
+# Wrapper to set the logging config for generations
+def update_from_dict(options_dict: Dict[str, bool]):
+    global config
+
+    # Force bools on the dict
+    for value in options_dict.values():
+        if value is None:
+            value = False
+
+    config = LogConfig.parse_obj(options_dict)
+
+def broadcast_status():
+    enabled = []
+    if config.prompt:
+        enabled.append("prompts")
+
+    if config.generation_params:
+        enabled.append("generation params")
+
+    if len(enabled) > 0:
+        print("Generation logging is enabled for: " + ", ".join(enabled))
+    else:
+        print("Generation logging is disabled")
+
+# Logs generation parameters to console
+def log_generation_params(**kwargs):
+    if config.generation_params:
+        print(f"Generation options: {kwargs}\n")
+
+def log_prompt(prompt: str):
+    if config.prompt:
+        print(f"Prompt: {prompt if prompt else 'Empty'}\n")
+
+def log_response(response: str):
+    if config.prompt:
+        print(f"Response: {response if response else 'Empty'}\n")

main.py

@@ -1,6 +1,7 @@
 import uvicorn
 import yaml
 import pathlib
+import gen_logging
 from asyncio import CancelledError
 from auth import check_admin_key, check_api_key, load_auth_keys
 from fastapi import FastAPI, Request, HTTPException, Depends
@@ -81,7 +82,8 @@ async def get_current_model():
             rope_alpha = model_container.config.scale_alpha_value,
             max_seq_len = model_container.config.max_seq_len,
             prompt_template = unwrap(model_container.prompt_template, "auto")
-        )
+        ),
+        logging = gen_logging.config
     )
 
     if model_container.draft_config:
@@ -370,6 +372,13 @@ if __name__ == "__main__":
         )
         config = {}
 
+    # Override the generation log options if given
+    log_config = unwrap(config.get("logging"), {})
+    if log_config:
+        gen_logging.update_from_dict(log_config)
+
+    gen_logging.broadcast_status()
+
     # If an initial model name is specified, create a container and load the model
     model_config = unwrap(config.get("model"), {})
     if "model_name" in model_config:

model.py

@@ -14,6 +14,7 @@ from exllamav2.generator import(
 )
 from typing import List, Optional, Union
 from utils import coalesce, unwrap
+from gen_logging import log_generation_params, log_prompt, log_response
 
 # Bytes to reserve on first device when loading with auto split
 auto_split_reserve_bytes = 96 * 1024**2
@@ -351,13 +352,6 @@ class ModelContainer:
         stop_conditions: List[Union[str, int]] = unwrap(kwargs.get("stop"), [])
         ban_eos_token = unwrap(kwargs.get("ban_eos_token"), False)
 
-
-        # Ban the EOS token if specified. If not, append to stop conditions as well.
-        if ban_eos_token:
-            gen_settings.disallow_tokens(self.tokenizer, [self.tokenizer.eos_token_id])
-        else:
-            stop_conditions.append(self.tokenizer.eos_token_id)
-
         # Override sampler settings for temp = 0
         if gen_settings.temperature == 0:
             gen_settings.temperature = 1.0
@@ -365,7 +359,25 @@ class ModelContainer:
             gen_settings.top_p = 0
             gen_settings.typical = 0
 
-        # Stop conditions           
+        # Log generation options to console
+        log_generation_params(
+            **vars(gen_settings),
+            token_healing = token_healing,
+            max_tokens = max_tokens,
+            stop_conditions = stop_conditions
+        )
+
+        # Log prompt to console
+        log_prompt(prompt)
+
+        # Ban the EOS token if specified. If not, append to stop conditions as well.
+        # Set this below logging to avoid polluting the stop strings array
+        if ban_eos_token:
+            gen_settings.disallow_tokens(self.tokenizer, [self.tokenizer.eos_token_id])
+        else:
+            stop_conditions.append(self.tokenizer.eos_token_id)
+
+        # Stop conditions
         self.generator.set_stop_conditions(stop_conditions)
 
         # Tokenized context
@@ -430,9 +442,12 @@ class ModelContainer:
 
             if eos or generated_tokens == max_tokens: break
 
+        # Print response
+        log_response(full_response)
+
         elapsed_time = last_chunk_time - start_time
 
-        initial_response = f"Response: {generated_tokens} tokens generated in {round(elapsed_time, 2)} seconds"
+        initial_response = f"Metrics: {generated_tokens} tokens generated in {round(elapsed_time, 2)} seconds"
         itemization = []
         extra_parts = []