metadata

tags:
  - generated_from_trainer
datasets:
  - toy_graph
metrics:
  - accuracy
model-index:
  - name: output_toy
    results:
      - task:
          name: Causal Language Modeling
          type: text-generation
        dataset:
          name: toy_graph
          type: toy_graph
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.4525254617525837

output_toy

This model is a fine-tuned version of toy/model on the toy_graph dataset. It achieves the following results on the evaluation set:

Loss: 1.2691
Accuracy: 0.4525
Transition Accuracy: 0.5634
First Transition Accuracy: 0.88
Multicode K: 1
Dead Code Fraction/layer0: 0.9969
Mse/layer0: 220380.4595
Input Norm/layer0: 333.7717
Output Norm/layer0: 12.9360
Dead Code Fraction/layer1: 0.9535
Mse/layer1: 132.7843
Input Norm/layer1: 6.5450
Output Norm/layer1: 13.1449
Dead Code Fraction/layer2: 0.9349
Mse/layer2: 365.9396
Input Norm/layer2: 6.1370
Output Norm/layer2: 18.3248
Dead Code Fraction/layer3: 0.9819
Mse/layer3: 415.9804
Input Norm/layer3: 7.4097
Output Norm/layer3: 18.4665

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.001
train_batch_size: 1024
eval_batch_size: 512
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: constant
training_steps: 20000

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Transition Accuracy	First Transition Accuracy	Multicode K	Dead Code Fraction/layer0	Dead Code Fraction/layer1	Dead Code Fraction/layer2	Dead Code Fraction/layer3
2.2465	0.03	500	1.8386	0.3565	0.3555	0.31	1	1.0	1.0	1.0	1.0
1.5981	0.05	1000	1.4652	0.4204	0.5015	0.58	1	1.0	1.0	1.0	1.0
1.3928	0.07	1500	1.3541	0.4378	0.555	0.79	1	1.0	1.0	1.0	1.0
1.3405	0.1	2000	1.3264	0.4427	0.5756	0.82	1	1.0	1.0	1.0	1.0
1.3189	0.12	2500	1.3187	0.4446	0.5576	0.86	1	1.0	1.0	1.0	1.0
1.308	0.15	3000	1.3064	0.4468	0.5573	0.82	1	1.0	1.0	1.0	1.0
1.3009	0.17	3500	1.2963	0.4493	0.5763	0.87	1	1.0	1.0	1.0	1.0
1.2965	0.2	4000	1.2922	0.4494	0.5677	0.9	1	1.0	1.0	1.0	1.0
1.2919	0.23	4500	1.2880	0.4499	0.5821	0.91	1	1.0	1.0	1.0	1.0
1.2889	0.25	5000	1.2856	0.4501	0.56	0.9	1	1.0	1.0	1.0	1.0
1.2855	0.28	5500	1.2816	0.4503	0.6016	0.9	1	1.0	1.0	1.0	1.0
1.2828	0.3	6000	1.2844	0.4502	0.5734	0.87	1	1.0	1.0	1.0	1.0
1.2805	0.33	6500	1.2777	0.4516	0.6084	0.95	1	1.0	1.0	1.0	1.0
1.2793	0.35	7000	1.2796	0.4511	0.5681	0.93	1	1.0	1.0	1.0	1.0
1.2785	0.38	7500	1.2748	0.4519	0.5919	0.95	1	1.0	1.0	1.0	1.0
1.2764	0.4	8000	1.2767	0.4518	0.5760	0.9	1	1.0	1.0	1.0	1.0
1.2763	0.42	8500	1.2801	0.4507	0.5827	0.94	1	1.0	1.0	1.0	1.0
1.2755	0.45	9000	1.2755	0.4516	0.5765	0.9	1	1.0	1.0	1.0	1.0
1.2746	0.47	9500	1.2736	0.4523	0.5865	0.9	1	1.0	1.0	1.0	1.0
1.2734	0.5	10000	1.2740	0.4519	0.5779	0.91	1	1.0	1.0	1.0	1.0
1.2732	0.53	10500	1.2744	0.4516	0.5879	0.89	1	1.0	1.0	1.0	1.0
1.2723	0.55	11000	1.2690	0.4525	0.5811	0.89	1	1.0	1.0	1.0	1.0
1.2712	0.57	11500	1.2705	0.4526	0.5779	0.93	1	1.0	1.0	1.0	1.0
1.2716	0.6	12000	1.2701	0.4527	0.5760	0.89	1	1.0	1.0	1.0	1.0
1.2708	0.62	12500	1.2716	0.4522	0.5485	0.95	1	1.0	1.0	1.0	1.0
1.2705	0.65	13000	1.2676	0.4529	0.5734	0.93	1	1.0	1.0	1.0	1.0
1.2696	0.68	13500	1.2717	0.4519	0.5994	0.91	1	1.0	1.0	1.0	1.0
1.2687	0.7	14000	1.2687	0.4524	0.5756	0.9	1	1.0	1.0	1.0	1.0
1.2685	0.72	14500	1.2709	0.4521	0.6127	0.89	1	1.0	1.0	1.0	1.0
1.2685	0.75	15000	1.2706	0.4519	0.5873	0.91	1	1.0	1.0	1.0	1.0
1.2675	0.78	15500	1.2691	0.4527	0.6365	0.96	1	1.0	1.0	1.0	1.0
1.2677	0.8	16000	1.2686	0.4526	0.5589	0.93	1	1.0	1.0	1.0	1.0
1.2676	0.82	16500	1.2639	0.4529	0.5940	0.89	1	1.0	1.0	1.0	1.0
1.2662	0.85	17000	1.2655	0.4530	0.5955	0.94	1	1.0	1.0	1.0	1.0
1.2666	0.88	17500	1.2636	0.4526	0.6013	0.96	1	1.0	1.0	1.0	1.0
1.2664	0.9	18000	1.2681	0.4526	0.6034	0.96	1	1.0	1.0	1.0	1.0
1.266	0.93	18500	1.2624	0.4527	0.5839	0.88	1	1.0	1.0	1.0	1.0
1.2653	0.95	19000	1.2688	0.4519	0.5837	0.92	1	1.0	1.0	1.0	1.0
1.2654	0.97	19500	1.2619	0.4534	0.5973	0.92	1	1.0	1.0	1.0	1.0
1.2649	1.0	20000	1.2647	0.4525	0.59	0.93	1	1.0	1.0	1.0	1.0

Framework versions

Transformers 4.28.1
Pytorch 2.0.1+cu117
Datasets 2.12.0
Tokenizers 0.13.3