Add provider cerebras, and update Mistral model availability

pkelaita · Dec 10, 2024 · 7e71b5c · 7e71b5c
1 parent a771834
commit 7e71b5c
Show file tree

Hide file tree

Showing 4 changed files with 47 additions and 28 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -6,9 +6,25 @@ _Current version: 0.0.36_
 
 ### 0.0.37 - December 9, 2024
 
+> [!CAUTION]
+> This release has _significant_ breaking changes! Please read the changelog carefully.
+
 #### Added
 
-- Support for [Cerebras](https://www.cerebras.net/) via [Groq](https://www.groq.com/).
+- Support for provider [Cerebras](https://cerebras.ai/), offering `llama-3.1-8b` and `llama-3.1-70b`.
+- Support for Mistral's `mistral-small`, `ministral-8b`, and `ministral-3b` models via La Plateforme.
+
+#### Changed
+
+- `mistral-large-2` has been renamed to `mistral-large`, to keep up with Mistral's naming scheme. **This is a breaking change!!!** Calls to `mistral-large-2` will fail.
+
+#### Removed
+
+- `mixtral-8x22b`, `mixtral-8x7b`, and `mistral-7b` are no longer available from provider Mistral as they have been [deprecated](https://docs.mistral.ai/getting-started/models/models_overview/). **This is a breaking change!!!** Calls to `mixtral-8x7b` and `mistral-7b` will fail, and calls to `mixtral-8x22b` via provider Mistral will fail.
+
+[!NOTE]
+
+> `mixtral-8x22b` is still available via Groq.
 
 ### 0.0.36 - November 21, 2024
 

diff --git a/README.md b/README.md
@@ -1,12 +1,12 @@
 # L2M2: A Simple Python LLM Manager 💬👍
 
-[![Tests](https://github.com/pkelaita/l2m2/actions/workflows/tests.yml/badge.svg?timestamp=1733805961)](https://github.com/pkelaita/l2m2/actions/workflows/tests.yml) [![codecov](https://codecov.io/github/pkelaita/l2m2/graph/badge.svg?token=UWIB0L9PR8)](https://codecov.io/github/pkelaita/l2m2) [![PyPI version](https://badge.fury.io/py/l2m2.svg?timestamp=1733805961)](https://badge.fury.io/py/l2m2)
+[![Tests](https://github.com/pkelaita/l2m2/actions/workflows/tests.yml/badge.svg?timestamp=1733808328)](https://github.com/pkelaita/l2m2/actions/workflows/tests.yml) [![codecov](https://codecov.io/github/pkelaita/l2m2/graph/badge.svg?token=UWIB0L9PR8)](https://codecov.io/github/pkelaita/l2m2) [![PyPI version](https://badge.fury.io/py/l2m2.svg?timestamp=1733808328)](https://badge.fury.io/py/l2m2)
 
 **L2M2** ("LLM Manager" &rarr; "LLMM" &rarr; "L2M2") is a tiny and very simple LLM manager for Python that exposes lots of models through a unified API. This is useful for evaluation, demos, production applications etc. that need to easily be model-agnostic.
 
 ### Features
 
-- <!--start-count-->26<!--end-count--> supported models (see below) – regularly updated and with more on the way.
+- <!--start-count-->27<!--end-count--> supported models (see below) – regularly updated and with more on the way.
 - Session chat memory – even across multiple models or with concurrent memory streams.
 - JSON mode
 - Prompt loading tools
@@ -38,10 +38,11 @@ L2M2 currently supports the following models:
 | `claude-3-haiku`    | [Anthropic](https://www.anthropic.com/api)                         | `claude-3-haiku-20240307`                           |
 | `command-r`         | [Cohere](https://docs.cohere.com/)                                 | `command-r`                                         |
 | `command-r-plus`    | [Cohere](https://docs.cohere.com/)                                 | `command-r-plus`                                    |
-| `mistral-large-2`   | [Mistral](https://mistral.ai/)                                     | `mistral-large-latest`                              |
-| `mixtral-8x22b`     | [Mistral](https://mistral.ai/)                                     | `open-mixtral-8x22b`                                |
-| `mixtral-8x7b`      | [Mistral](https://mistral.ai/), [Groq](https://wow.groq.com/)      | `open-mixtral-8x7b`, `mixtral-8x7b-32768`           |
-| `mistral-7b`        | [Mistral](https://mistral.ai/)                                     | `open-mistral-7b`                                   |
+| `mistral-large`     | [Mistral](https://mistral.ai/)                                     | `mistral-large-latest`                              |
+| `ministral-3b`      | [Mistral](https://mistral.ai/)                                     | `ministral-3b-latest`                               |
+| `ministral-8b`      | [Mistral](https://mistral.ai/)                                     | `ministral-8b-latest`                               |
+| `mistral-small`     | [Mistral](https://mistral.ai/)                                     | `mistral-small-latest`                              |
+| `mixtral-8x7b`      | [Groq](https://wow.groq.com/)                                      | `mixtral-8x7b-32768`                                |
 | `gemma-7b`          | [Groq](https://wow.groq.com/)                                      | `gemma-7b-it`                                       |
 | `gemma-2-9b`        | [Groq](https://wow.groq.com/)                                      | `gemma2-9b-it`                                      |
 | `llama-3-8b`        | [Groq](https://wow.groq.com/), [Replicate](https://replicate.com/) | `llama3-8b-8192`, `meta/meta-llama-3-8b-instruct`   |
@@ -512,10 +513,10 @@ The following models natively support JSON mode via the given provider:
 - `gpt-4-turbo` (via Openai)
 - `gpt-3.5-turbo` (via Openai)
 - `gemini-1.5-pro` (via Google)
-- `mistral-large-2` (via Mistral)
-- `mixtral-8x22b` (via Mistral)
-- `mixtral-8x7b` (via Mistral)
-- `mistral-7b` (via Mistral)
+- `mistral-large` (via Mistral)
+- `ministral-3b` (via Mistral)
+- `ministral-8b` (via Mistral)
+- `mistral-small` (via Mistral)
 
 <!--end-json-native-->
 

diff --git a/l2m2/model_info.py b/l2m2/model_info.py
@@ -335,7 +335,7 @@ class ModelEntry(TypedDict):
             "extras": {},
         },
     },
-    "mistral-large-2": {
+    "mistral-large": {
         "mistral": {
             "model_id": "mistral-large-latest",
             "params": {
@@ -351,9 +351,9 @@ class ModelEntry(TypedDict):
             "extras": {"json_mode_arg": {"response_format": {"type": "json_object"}}},
         },
     },
-    "mixtral-8x22b": {
+    "ministral-3b": {
         "mistral": {
-            "model_id": "open-mixtral-8x22b",
+            "model_id": "ministral-3b-latest",
             "params": {
                 "temperature": {
                     "default": PROVIDER_DEFAULT,
@@ -367,9 +367,9 @@ class ModelEntry(TypedDict):
             "extras": {"json_mode_arg": {"response_format": {"type": "json_object"}}},
         },
     },
-    "mixtral-8x7b": {
+    "ministral-8b": {
         "mistral": {
-            "model_id": "open-mixtral-8x7b",
+            "model_id": "ministral-8b-latest",
             "params": {
                 "temperature": {
                     "default": PROVIDER_DEFAULT,
@@ -382,35 +382,37 @@ class ModelEntry(TypedDict):
             },
             "extras": {"json_mode_arg": {"response_format": {"type": "json_object"}}},
         },
-        "groq": {
-            "model_id": "mixtral-8x7b-32768",
+    },
+    "mistral-small": {
+        "mistral": {
+            "model_id": "mistral-small-latest",
             "params": {
                 "temperature": {
                     "default": PROVIDER_DEFAULT,
-                    "max": 2.0,
+                    "max": 1.0,
                 },
                 "max_tokens": {
                     "default": PROVIDER_DEFAULT,
-                    "max": 2**16 - 1,
+                    "max": INF,
                 },
             },
-            "extras": {},
+            "extras": {"json_mode_arg": {"response_format": {"type": "json_object"}}},
         },
     },
-    "mistral-7b": {
-        "mistral": {
-            "model_id": "open-mistral-7b",
+    "mixtral-8x7b": {
+        "groq": {
+            "model_id": "mixtral-8x7b-32768",
             "params": {
                 "temperature": {
                     "default": PROVIDER_DEFAULT,
-                    "max": 1.0,
+                    "max": 2.0,
                 },
                 "max_tokens": {
                     "default": PROVIDER_DEFAULT,
-                    "max": INF,
+                    "max": 2**16 - 1,
                 },
             },
-            "extras": {"json_mode_arg": {"response_format": {"type": "json_object"}}},
+            "extras": {},
         },
     },
     "gemma-7b": {

diff --git a/tests/l2m2/client/test_base_llm_client.py b/tests/l2m2/client/test_base_llm_client.py
@@ -304,7 +304,7 @@ async def test_call_mistral(mock_get_extra_message, mock_llm_post, llm_client):
     mock_get_extra_message.return_value = "extra message"
     mock_return_value = {"choices": [{"message": {"content": "response"}}]}
     mock_llm_post.return_value = mock_return_value
-    await _generic_test_call(llm_client, "mistral", "mixtral-8x7b")
+    await _generic_test_call(llm_client, "mistral", "mistral-large")
 
 
 @pytest.mark.asyncio