Training in progress, step 38

README.md

@@ -1,58 +1,5 @@
----
-base_model: google/gemma-3-27b-it
-library_name: transformers
-model_name: finetune-demo-lora
-tags:
-- generated_from_trainer
-- trl
-- sft
-licence: license
----
+# Baseten Training Demo 
 
-# Model Card for finetune-demo-lora
+This is a demo of how to finetune a model using Baseten Training.
 
-This model is a fine-tuned version of [google/gemma-3-27b-it](https://huggingface.co/google/gemma-3-27b-it).
-It has been trained using [TRL](https://github.com/huggingface/trl).
-
-## Quick start
-
-```python
-from transformers import pipeline
-
-question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
-generator = pipeline("text-generation", model="rayraycano/finetune-demo-lora", device="cuda")
-output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
-print(output["generated_text"])
-```
-
-## Training procedure
-
-[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/philipkiely-baseten/baseten-finetune-demo/runs/y2yddifv) 
-
-
-This model was trained with SFT.
-
-### Framework versions
-
-- TRL: 0.15.1
-- Transformers: 4.50.0
-- Pytorch: 2.6.0+cu124
-- Datasets: 3.4.1
-- Tokenizers: 0.21.1
-
-## Citations
-
-
-
-Cite TRL as:
-    
-```bibtex
-@misc{vonwerra2022trl,
-	title        = {{TRL: Transformer Reinforcement Learning}},
-	author       = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
-	year         = 2020,
-	journal      = {GitHub repository},
-	publisher    = {GitHub},
-	howpublished = {\url{https://github.com/huggingface/trl}}
-}
-```
+See the [Get Started](https://docs.baseten.com/get-started) guide to learn more about how to use Baseten Training.

adapter_config.json

@@ -24,13 +24,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
     "gate_proj",
     "q_proj",
-    "v_proj",
+    "o_proj",
+    "k_proj",
     "down_proj",
     "up_proj",
-    "o_proj"
+    "v_proj"
   ],
   "task_type": "CAUSAL_LM",
   "trainable_token_indices": null,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:450fd25cba6e995f2066bb92b69bae186c5243fa7b9131e3c3d7857636730bc2
+oid sha256:0866d11ec1b3e6c3d52e42ce0f9466c229d211d6ac1c09da21e95187ecc836ac
 size 113632496

config.py

@@ -18,9 +18,9 @@ runtime = definitions.Runtime(
         "OUTPUT_LORA_REPO_ID": "rayraycano/finetune-demo-lora", # TODO: your HF Repo ID
     },
     enable_cache=True,
-    checkpointing_config=definitions.CheckpointingConfig(
-        enabled=False,
-    ),
+    # checkpointing_config=definitions.CheckpointingConfig(
+    #     enabled=True,
+    # ),
 )
 
 """
@@ -32,7 +32,7 @@ compute = definitions.Compute(
         accelerator=truss_config.Accelerator.H200,
         count=8,
     ),
-    # node_count=2,
+    node_count=2,
 )
 
 """

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1cc0791c494bc9424165d7d11c36192341830d3811c11cbb5ba458a3547e56ac
-size 6904
+oid sha256:845e5541434556dc2b5a7fd6819ebe6e1e49ee5b3a7e173ee184a0043de929f1
+size 6840