Fixing nits

Signed-off-by: Vladimir Suvorov <[email protected]>

Author: mydatascience

…ext/examples/grpo_llama3_1_8b_demo.ipynb …/MaxText/examples/grpo_llama3_demo.ipynbsrc/MaxText/examples/grpo_llama3_1_8b_demo.ipynb renamed to src/MaxText/examples/grpo_llama3_demo.ipynb src/MaxText/examples/grpo_llama3_1_8b_demo.ipynb renamed to src/MaxText/examples/grpo_llama3_demo.ipynb

@@ -19,15 +19,7 @@
 set -e
 set -x
 
-<<<<<<< HEAD
-<<<<<<< HEAD
 uv pip uninstall jax jaxlib libtpu
-=======
-yes | uv pip uninstall jax jaxlib libtpu
->>>>>>> 08e2cdb19 (Add GSPO and small fixes)
-=======
-uv pip uninstall jax jaxlib libtpu
->>>>>>> de8d3d1ce (Fix)
 
 uv pip install aiohttp==3.12.15
 

src/MaxText/examples/install_tunix_vllm_requirement.sh

@@ -17,34 +17,7 @@
     "\n",
     "This notebook can run on **TPU v5e-8** or **v5p-8**\n",
     "\n",
-    "This notebook demonstrates how to perform Supervised Fine-Tuning (SFT) on Llama3.1-8B-Instruct using the Hugging Face ultrachat_200k dataset with MaxText and Tunix integration for efficient training.\n",
-    "\n",
-    "## Dataset Overview\n",
-    "\n",
-    "**Dataset Link:** https://huggingface.co/datasets/HuggingFaceH4/ultrachat_200k\n",
-    "\n",
-    "### Dataset Information:\n",
-    "- **Name:** HuggingFaceH4/ultrachat_200k\n",
-    "- **Type:** Supervised Fine-Tuning dataset\n",
-    "- **Size:** ~200k conversations\n",
-    "- **Format:** Chat conversations with human-AI pairs\n",
-    "- **Splits:** train_sft, test_sft\n",
-    "- **Data columns:** ['messages']\n",
-    "\n",
-    "### Dataset Structure:\n",
-    "Each example contains a 'messages' field with:\n",
-    "- **role:** 'user' or 'assistant'\n",
-    "- **content:** The actual message text\n",
-    "\n",
-    "### Example data format:\n",
-    "```json\n",
-    "{\n",
-    "  \"messages\": [\n",
-    "    {\"role\": \"user\", \"content\": \"What is the capital of France?\"},\n",
-    "    {\"role\": \"assistant\", \"content\": \"The capital of France is Paris.\"}\n",
-    "  ]\n",
-    "}\n",
-    "```\n"
+    "This notebook demonstrates how to perform Supervised Fine-Tuning (SFT) on Llama3.1-8B-Instruct using the Hugging Face ultrachat_200k dataset with MaxText and Tunix integration for efficient training.\n"
    ]
   },
   {
@@ -84,7 +57,7 @@
     "\n",
     "**Follow these steps to store your token:**\n",
     "\n",
-    "Just put your token in the line below"
+    "Just put your token in the cell below"
    ]
   },
   {
@@ -100,7 +73,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Install maxtext and dependencies\n",
+    "### Install\n",
     "\n",
     "https://maxtext.readthedocs.io/en/latest/tutorials/sft.html#install-dependencies"
    ]
@@ -118,7 +91,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!cd ~/maxtext/src/ #  This is the path to the maxtext folder"
+    "!cd ~/maxtext/src/ #  make sure we are in the right directory"
    ]
   },
   {