Examples: add options to save or push model (huggingface#1159)

BenjaminBossan · Nov 30, 2023 · 00dc6f4 · 00dc6f4
1 parent 570e992
commit 00dc6f4
Show file tree

Hide file tree

Showing 6 changed files with 125 additions and 20 deletions.
diff --git a/examples/causal_language_modeling/peft_lora_clm_accelerate_big_model_inference.ipynb b/examples/causal_language_modeling/peft_lora_clm_accelerate_big_model_inference.ipynb
@@ -210,6 +210,23 @@
     "print(next(iter(test_dataloader)))"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "42b14a11",
+   "metadata": {},
+   "source": [
+    "You can load model from hub or local\n",
+    "\n",
+    "- Load model from Hugging Face Hub, you can change to your own model id\n",
+    "```python\n",
+    "peft_model_id = \"username/twitter_complaints_bigscience_bloomz-7b1_LORA_CAUSAL_LM\"\n",
+    "```\n",
+    "- Or load model form local\n",
+    "```python\n",
+    "peft_model_id = \"twitter_complaints_bigscience_bloomz-7b1_LORA_CAUSAL_LM\"\n",
+    "```"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 5,
@@ -244,7 +261,6 @@
     "\n",
     "max_memory = {0: \"1GIB\", 1: \"1GIB\", 2: \"2GIB\", 3: \"10GIB\", \"cpu\": \"30GB\"}\n",
     "peft_model_id = \"smangrul/twitter_complaints_bigscience_bloomz-7b1_LORA_CAUSAL_LM\"\n",
-    "\n",
     "config = PeftConfig.from_pretrained(peft_model_id)\n",
     "model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path, device_map=\"auto\", max_memory=max_memory)\n",
     "model = PeftModel.from_pretrained(model, peft_model_id, device_map=\"auto\", max_memory=max_memory)"

diff --git a/examples/causal_language_modeling/peft_lora_clm_accelerate_ds_zero3_offload.py b/examples/causal_language_modeling/peft_lora_clm_accelerate_ds_zero3_offload.py
@@ -349,12 +349,21 @@ def test_preprocess_function(examples):
         pred_df.to_csv(f"data/{dataset_name}/predictions.csv", index=False)
 
     accelerator.wait_for_everyone()
-    model.push_to_hub(
-        "smangrul/"
-        + f"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}".replace("/", "_"),
-        state_dict=accelerator.get_state_dict(model),
-        use_auth_token=True,
+    # Option1: Pushing the model to Hugging Face Hub
+    # model.push_to_hub(
+    #     f"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}".replace("/", "_"),
+    #     token = "hf_..."
+    # )
+    # token (`bool` or `str`, *optional*):
+    #     `token` is to be used for HTTP Bearer authorization when accessing remote files. If `True`, will use the token generated
+    #     when running `huggingface-cli login` (stored in `~/.huggingface`). Will default to `True` if `repo_url`
+    #     is not specified.
+    #     Or you can get your token from https://huggingface.co/settings/token
+    # Option2: Saving the model locally
+    peft_model_id = f"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}".replace(
+        "/", "_"
     )
+    model.save_pretrained(peft_model_id)
     accelerator.wait_for_everyone()
 
 

diff --git a/examples/causal_language_modeling/peft_prefix_tuning_clm.ipynb b/examples/causal_language_modeling/peft_prefix_tuning_clm.ipynb
@@ -1228,6 +1228,33 @@
     "    print(tokenizer.batch_decode(outputs.detach().cpu().numpy(), skip_special_tokens=True))"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "0e21c49b",
+   "metadata": {},
+   "source": [
+    "You can push model to hub or save model locally. \n",
+    "\n",
+    "- Option1: Pushing the model to Hugging Face Hub\n",
+    "```python\n",
+    "model.push_to_hub(\n",
+    "    f\"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\".replace(\"/\", \"_\"),\n",
+    "    token = \"hf_...\"\n",
+    ")\n",
+    "```\n",
+    "token (`bool` or `str`, *optional*):\n",
+    "    `token` is to be used for HTTP Bearer authorization when accessing remote files. If `True`, will use the token generated\n",
+    "    when running `huggingface-cli login` (stored in `~/.huggingface`). Will default to `True` if `repo_url`\n",
+    "    is not specified.\n",
+    "    Or you can get your token from https://huggingface.co/settings/token\n",
+    "```\n",
+    "- Or save model locally\n",
+    "```python\n",
+    "peft_model_id = f\"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\".replace(\"/\", \"_\")\n",
+    "model.save_pretrained(peft_model_id)\n",
+    "```"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 16,
@@ -1236,7 +1263,9 @@
    "outputs": [],
    "source": [
     "# saving model\n",
-    "peft_model_id = f\"{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\"\n",
+    "peft_model_id = f\"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\".replace(\n",
+    "    \"/\", \"_\"\n",
+    ")\n",
     "model.save_pretrained(peft_model_id)"
    ]
   },
@@ -1260,7 +1289,9 @@
    "source": [
     "from peft import PeftModel, PeftConfig\n",
     "\n",
-    "peft_model_id = f\"{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\"\n",
+    "peft_model_id = f\"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\".replace(\n",
+    "    \"/\", \"_\"\n",
+    ")\n",
     "\n",
     "config = PeftConfig.from_pretrained(peft_model_id)\n",
     "model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path)\n",

diff --git a/examples/causal_language_modeling/peft_prompt_tuning_clm.ipynb b/examples/causal_language_modeling/peft_prompt_tuning_clm.ipynb
@@ -1072,6 +1072,33 @@
     "    print(tokenizer.batch_decode(outputs.detach().cpu().numpy(), skip_special_tokens=True))"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "c8f35152",
+   "metadata": {},
+   "source": [
+    "You can push model to hub or save model locally. \n",
+    "\n",
+    "- Option1: Pushing the model to Hugging Face Hub\n",
+    "```python\n",
+    "model.push_to_hub(\n",
+    "    f\"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\".replace(\"/\", \"_\"),\n",
+    "    token = \"hf_...\"\n",
+    ")\n",
+    "```\n",
+    "token (`bool` or `str`, *optional*):\n",
+    "    `token` is to be used for HTTP Bearer authorization when accessing remote files. If `True`, will use the token generated\n",
+    "    when running `huggingface-cli login` (stored in `~/.huggingface`). Will default to `True` if `repo_url`\n",
+    "    is not specified.\n",
+    "    Or you can get your token from https://huggingface.co/settings/token\n",
+    "```\n",
+    "- Or save model locally\n",
+    "```python\n",
+    "peft_model_id = f\"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\".replace(\"/\", \"_\")\n",
+    "model.save_pretrained(peft_model_id)\n",
+    "```"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 12,
@@ -1080,7 +1107,9 @@
    "outputs": [],
    "source": [
     "# saving model\n",
-    "peft_model_id = f\"{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\"\n",
+    "peft_model_id = f\"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\".replace(\n",
+    "    \"/\", \"_\"\n",
+    ")\n",
     "model.save_pretrained(peft_model_id)"
    ]
   },
@@ -1116,7 +1145,9 @@
    "source": [
     "from peft import PeftModel, PeftConfig\n",
     "\n",
-    "peft_model_id = f\"{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\"\n",
+    "peft_model_id = f\"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}\".replace(\n",
+    "    \"/\", \"_\"\n",
+    ")\n",
     "\n",
     "config = PeftConfig.from_pretrained(peft_model_id)\n",
     "model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path)\n",

diff --git a/examples/conditional_generation/peft_lora_seq2seq_accelerate_ds_zero3_offload.py b/examples/conditional_generation/peft_lora_seq2seq_accelerate_ds_zero3_offload.py
@@ -298,12 +298,22 @@ def collate_fn(examples):
         pred_df.to_csv(f"data/{dataset_name}/predictions.csv", index=False)
 
     accelerator.wait_for_everyone()
-    model.push_to_hub(
-        "smangrul/"
-        + f"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}".replace("/", "_"),
-        state_dict=accelerator.get_state_dict(model),
-        use_auth_token=True,
+    # Option1: Pushing the model to Hugging Face Hub
+    # model.push_to_hub(
+    #     f"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}".replace("/", "_"),
+    #     token = "hf_..."
+    # )
+    # token (`bool` or `str`, *optional*):
+    #     `token` is to be used for HTTP Bearer authorization when accessing remote files. If `True`, will use the token generated
+    #     when running `huggingface-cli login` (stored in `~/.huggingface`). Will default to `True` if `repo_url`
+    #     is not specified.
+    #     Or you can get your token from https://huggingface.co/settings/token
+
+    # Option2: Saving the model locally
+    peft_model_id = f"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}".replace(
+        "/", "_"
     )
+    model.save_pretrained(peft_model_id)
     accelerator.wait_for_everyone()
 
 

diff --git a/examples/conditional_generation/peft_lora_seq2seq_accelerate_fsdp.py b/examples/conditional_generation/peft_lora_seq2seq_accelerate_fsdp.py
@@ -125,11 +125,19 @@ def preprocess_function(examples):
         accelerator.print(f"{eval_preds[:10]=}")
         accelerator.print(f"{dataset['validation'][label_column][:10]=}")
         accelerator.wait_for_everyone()
-        model.push_to_hub(
-            "smangrul/" + f"{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}".replace("/", "_"),
-            state_dict=accelerator.get_state_dict(model),
-            use_auth_token=True,
-        )
+        # Option1: Pushing the model to Hugging Face Hub
+        # model.push_to_hub(
+        #     f"{dataset_name}_{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}".replace("/", "_"),
+        #     token = "hf_..."
+        # )
+        # token (`bool` or `str`, *optional*):
+        #     `token` is to be used for HTTP Bearer authorization when accessing remote files. If `True`, will use the token generated
+        #     when running `huggingface-cli login` (stored in `~/.huggingface`). Will default to `True` if `repo_url`
+        #     is not specified.
+        #     Or you can get your token from https://huggingface.co/settings/token
+        # Option2: Saving the model locally
+        peft_model_id = f"{model_name_or_path}_{peft_config.peft_type}_{peft_config.task_type}".replace("/", "_")
+        model.save_pretrained(peft_model_id)
         accelerator.wait_for_everyone()