Refactor LLM monitor. Add the ability to use monitor as a context manager. Centralize OpenAI costs table

Author: gustavocidornelas

CHANGELOG.md

@@ -8,6 +8,7 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 ## Unreleased
 
 ### Added
+* Added the ability to use the `llm_monitor.OpenAIMonitor` as a context manager.
 * Added `openlayer_inference_pipeline_id` as an optional parameter to the `OpenAIMonitor`. This is an alternative to `openlayer_inference_pipeline_name` and `openlayer_inference_project_name` parameters for identifying the inference pipeline on the platform.
 * Added `monitor_output_only` as an argument to the OpenAI `llm_monitor`. If set to `True`, the monitor will only record the output of the model, and not the input.
 * Added `costColumnName` as an optional field in the config for LLM data.

openlayer/constants.py

@@ -34,3 +34,71 @@
     ),
     COLUMN_NAME_REGEX,
 ]
+# --------------------------- LLM usage costs table -------------------------- #
+# Last update: 2024-02-05
+OPENAI_COST_PER_TOKEN = {
+    "babbage-002": {
+        "input": 0.0004e-3,
+        "output": 0.0004e-3,
+    },
+    "davinci-002": {
+        "input": 0.002e-3,
+        "output": 0.002e-3,
+    },
+    "gpt-3.5-turbo": {
+        "input": 0.0005e-3,
+        "output": 0.0015e-3,
+    },
+    "gpt-3.5-turbo-0125": {
+        "input": 0.0005e-3,
+        "output": 0.0015e-3,
+    },
+    "gpt-3.5-turbo-0301": {
+        "input": 0.0015e-3,
+        "output": 0.002e-3,
+    },
+    "gpt-3.5-turbo-0613": {
+        "input": 0.0015e-3,
+        "output": 0.002e-3,
+    },
+    "gpt-3.5-turbo-1106": {
+        "input": 0.001e-3,
+        "output": 0.002e-3,
+    },
+    "gpt-3.5-turbo-16k-0613": {
+        "input": 0.003e-3,
+        "output": 0.004e-3,
+    },
+    "gpt-3.5-turbo-instruct": {
+        "input": 0.0015e-3,
+        "output": 0.002e-3,
+    },
+    "gpt-4": {
+        "input": 0.03e-3,
+        "output": 0.06e-3,
+    },
+    "gpt-4-0125-preview": {
+        "input": 0.01e-3,
+        "output": 0.03e-3,
+    },
+    "gpt-4-1106-preview": {
+        "input": 0.01e-3,
+        "output": 0.03e-3,
+    },
+    "gpt-4-0314": {
+        "input": 0.03e-3,
+        "output": 0.06e-3,
+    },
+    "gpt-4-1106-vision-preview": {
+        "input": 0.01e-3,
+        "output": 0.03e-3,
+    },
+    "gpt-4-32k": {
+        "input": 0.06e-3,
+        "output": 0.12e-3,
+    },
+    "gpt-4-32k-0314": {
+        "input": 0.06e-3,
+        "output": 0.12e-3,
+    },
+}