anlp-a3-mamba

Character-level cipher decryption model (mamba), ANLP Assignment 3 Part A.

Architecture: mamba, layers: 12

metric value
params 5284864
flops_fwd_1024 11238637568
test_char_accuracy 0.1738
test_word_accuracy 0.0035
test_precision 0.7286
test_recall 0.0427
test_f1 0.0220
test_precision_micro 0.1738
test_recall_micro 0.1738
test_f1_micro 0.1738
test_levenshtein_mean 734.5903
test_levenshtein_norm_mean 0.7099
test_cer 0.8262
test_exact_match 0.0137
test_length_match 0.9992
test_infer_seconds 35.6716
test_peak_mem_gb 2.1195
test_loss 2.5088
Downloads last month
52
Safetensors
Model size
5.28M params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support