Granite 3.1 (#40)

* feat: Update Ollama for 3.1 models Also add references to the MoE models and remove references to Granite Code Branch: Granite31 Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat: Update Replicate for the 3.1 models Replicate will host the 8b and 2b dense models and we will stop recommending Granite Code Branch: Granite31 Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat: Update WatsonX for the 3.1 models Add MoE and remove Granite Code Branch: Granite31 Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat: Update to 3.1 for Langchain LLMs Branch: Granite31 Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat: Update langchain embeddings to use granite embeddings Branch: Granite31 Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Fix watsonx model IDs Branch: Granite31 Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Default to 30m embedding model Branch: Granite31 Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> --------- Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
ibm-granite-community · Dec 18, 2024 · 0f452af · 0f452af
1 parent dfecc88
commit 0f452af
Show file tree

Hide file tree

Showing 5 changed files with 24 additions and 37 deletions.
diff --git a/recipes/Components/Langchain_Embeddings_Models.ipynb b/recipes/Components/Langchain_Embeddings_Models.ipynb
@@ -37,9 +37,8 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# Commented out to avoid downloading the model.\n",
-    "# from langchain_huggingface import HuggingFaceEmbeddings\n",
-    "# embeddings_model = HuggingFaceEmbeddings(model_name=\"all-MiniLM-L6-v2\")"
+    "from langchain_huggingface import HuggingFaceEmbeddings\n",
+    "embeddings_model = HuggingFaceEmbeddings(model_name=\"ibm-granite/granite-embedding-30m-english\")"
    ]
   },
   {

diff --git a/recipes/Components/Langchain_LLMs.ipynb b/recipes/Components/Langchain_LLMs.ipynb
@@ -52,7 +52,7 @@
     "from ibm_granite_community.notebook_utils import get_env_var\n",
     "\n",
     "model = Replicate(\n",
-    "    model=\"ibm-granite/granite-3.0-8b-instruct\",\n",
+    "    model=\"ibm-granite/granite-3.1-8b-instruct\",\n",
     "    replicate_api_token=get_env_var('REPLICATE_API_TOKEN'),\n",
     ")"
    ]
@@ -78,7 +78,7 @@
    "source": [
     "from langchain_ollama.llms import OllamaLLM\n",
     "\n",
-    "model = OllamaLLM(model=\"granite3-dense:8b\")"
+    "model = OllamaLLM(model=\"granite3.1-dense:8b\")"
    ]
   },
   {

diff --git a/recipes/Getting_Started/Getting_Started_with_Ollama.ipynb b/recipes/Getting_Started/Getting_Started_with_Ollama.ipynb
@@ -51,11 +51,10 @@
     "1. Pull down Granite models:\n",
     "\n",
     "    ```shell\n",
-    "    ollama pull granite3-dense:2b\n",
-    "    ollama pull granite3-dense:8b\n",
-    "    ollama pull granite-code:3b\n",
-    "    ollama pull granite-code:8b\n",
-    "    ollama pull granite-code:20b\n",
+    "    ollama pull granite3.1-dense:2b\n",
+    "    ollama pull granite3.1-dense:8b\n",
+    "    ollama pull granite3.1-moe:1b\n",
+    "    ollama pull granite3.1-moe:3b\n",
     "    ```"
    ]
   },
@@ -107,11 +106,10 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!ollama pull granite3-dense:2b\n",
-    "# !ollama pull granite3-dense:8b\n",
-    "# !ollama pull granite-code:3b\n",
-    "# !ollama pull granite-code:8b\n",
-    "# !ollama pull granite-code:20b"
+    "!ollama pull granite3.1-dense:2b\n",
+    "# !ollama pull granite3.1-dense:8b\n",
+    "# !ollama pull granite3.1-moe:1b\n",
+    "# !ollama pull granite3.1-moe:3b"
    ]
   },
   {
@@ -145,11 +143,10 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "model_id = \"granite3-dense:2b\"\n",
-    "# model_id = \"granite3-dense:8b\"\n",
-    "# model_id = \"granite-code:3b\"\n",
-    "# model_id = \"granite-code:8b\"\n",
-    "# model_id = \"granite-code:20b\""
+    "model_id = \"granite3.1-dense:2b\"\n",
+    "# model_id = \"granite3.1-dense:8b\"\n",
+    "# model_id = \"granite3.1-moe:1b\"\n",
+    "# model_id = \"granite3.1-moe:3b\""
    ]
   },
   {

diff --git a/recipes/Getting_Started/Getting_Started_with_Replicate.ipynb b/recipes/Getting_Started/Getting_Started_with_Replicate.ipynb
@@ -99,15 +99,9 @@
     "\n",
     "Granite models are available in the [`ibm-granite`](https://replicate.com/ibm-granite) org on Replicate.\n",
     "\n",
-    "Two Granite Code models:\n",
+    "`ibm-granite/granite-3.1-8b-instruct`\n",
     "\n",
-    "`ibm-granite/granite-8b-code-instruct-128k`\n",
-    "\n",
-    "`ibm-granite/granite-20b-code-instruct-8k`\n",
-    "\n",
-    "And a Granite language model:\n",
-    "\n",
-    "`ibm-granite/granite-3.0-8b-instruct`"
+    "`ibm-granite/granite-3.1-2b-instruct`"
    ]
   },
   {
@@ -129,7 +123,7 @@
     "from ibm_granite_community.notebook_utils import get_env_var\n",
     "\n",
     "model = Replicate(\n",
-    "    model=\"ibm-granite/granite-3.0-8b-instruct\",\n",
+    "    model=\"ibm-granite/granite-3.1-8b-instruct\",\n",
     "    replicate_api_token=get_env_var('REPLICATE_API_TOKEN'),\n",
     ")"
    ]

diff --git a/recipes/Getting_Started/Getting_Started_with_WatsonX.ipynb b/recipes/Getting_Started/Getting_Started_with_WatsonX.ipynb
@@ -119,9 +119,9 @@
     "\n",
     "The Granite models available on WatsonX are listed on the [Resource Hub](https://dataplatform.cloud.ibm.com/samples?context=wx&tab=foundation-model&query=granite). For more information about Granite models on WatsonX, see [Foundation models on watsonx.ai](https://www.ibm.com/products/watsonx-ai/foundation-models).\n",
     "\n",
-    "[`ibm/granite-8b-code-instruct`](https://dataplatform.cloud.ibm.com/wx/samples/models/ibm/granite-8b-code-instruct?context=wx)\n",
+    "[`ibm/granite-3-8b-instruct`](https://dataplatform.cloud.ibm.com/wx/samples/models/ibm/granite-3-8b-instruct?context=wx)\n",
     "\n",
-    "[`ibm/granite-3-8b-instruct`](https://dataplatform.cloud.ibm.com/wx/samples/models/ibm/granite-3-8b-instruct?context=wx)"
+    "[`ibm/granite-3-2b-instruct`](https://dataplatform.cloud.ibm.com/wx/samples/models/ibm/granite-3-2b-instruct?context=wx)"
    ]
   },
   {
@@ -131,11 +131,8 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# model_id = \"ibm/granite-3b-code-instruct\"\n",
-    "# model_id = \"ibm/granite-8b-code-instruct\"\n",
-    "# model_id = \"ibm/granite-20b-code-instruct\"\n",
-    "# model_id = \"ibm/granite-34b-code-instruct\"\n",
-    "model_id = \"ibm/granite-3-8b-instruct\""
+    "model_id = \"ibm/granite-3-8b-instruct\"\n",
+    "# model_id = \"ibm/granite-3-2b-instruct\""
    ]
   },
   {
@@ -158,7 +155,7 @@
     "from langchain_ibm import WatsonxLLM\n",
     "\n",
     "model = WatsonxLLM(\n",
-    "    model_id=model_id, \n",
+    "    model_id=model_id,\n",
     "    url= get_env_var(\"WATSONX_URL\"),\n",
     "    apikey=get_env_var(\"WATSONX_APIKEY\"),\n",
     "    project_id=get_env_var(\"WATSONX_PROJECT_ID\"),\n",