Update README.md

README.md

@@ -1,16 +1,67 @@
 ---
+language:
+- code
 license: bigcode-openrail-m
 datasets:
 - bigcode/the-stack-dedup
-- Vipitis/Shadertoys
+- Vipitis/Shadertoys-fine
 pipeline_tag: text-generation
 tags:
 - code
+- shader
+widget:
+- text: void mainImage( out vec4 fragColor, in vec2 fragCoord )
+  example_title: mainImage
+  group: Shadertoy
+model-index:
+- name: santacoder-finetuned-the-stack-glsl
+  results:
+  - task:
+      type: text-generation
+      name: ShaderEval
+    dataset:
+      type: Vipitis/Shadertoys-fine
+      name: Shadertoys-fine
+      config: return_completion
+      revision: 0.0.2
+    metrics:
+      - type: exact_match
+        value: 0.550
+        name: 300 samples, greedy decoding
+        verified: false
 ---
 
 [Santacoder](https://huggingface.co/bigcode/santacoder) finetuned on [Shadertoys](https://huggingface.co/datasets/Vipitis/Shadertoys) for 1000 steps with a batch size of 2 and full sequence length of 2048.
-Origianl finetuning script from found [here](https://github.com/loubnabnl/santacoder-finetuning), adapted version to follow (soon^^).
+adapted finetuning script found [here](./train.py)
+
+Try model in the [ShaderCoder](https://huggingface.co/spaces/Vipitis/ShaderCoder) demo space
+
+### Finetuning parameters
+```sh
+python3 train.py --model_path "bigcode/santacoder" \
+--dataset_name "Vipitis/Shadertoys" \
+--data_column "code" \
+--split "train" \
+--seq_length 2048 \
+--max_steps 1000 \
+--batch_size 2 \
+--gradient_accumulation_steps 4 \
+--learning_rate 5e-5 \
+--num_warmup_steps 100 \
+--eval_freq 100 \
+--save_freq 100 \
+--log_freq 1 \
+--output_dir "checkpoint_dir" \
+--no_fp16
+
+
+```
+
+Main purpose of this model is to explore if finetuning models improves performance on [ShaderEval](https://huggingface.co/spaces/Vipitis/ShaderEval), which reached 0.550 with 300 samples.
+
+### Disclaimer
 
-Main purpose of this model is to explore if finetuning models improves performance on [ShaderEval](https://huggingface.co/spaces/Vipitis/ShaderEval), results to follow (sooner).
+While the train/test split is held out, there is a lot of data contamination. The model results can't be trusted for this simple benchmark.
+Better tasks for the benchmark will be developed and tested against these models.
 
 License carried over from model, however training data has an undefied license. Check details in [Shadertoys](https://huggingface.co/datasets/Vipitis/Shadertoys).