Upload folder using huggingface_hub

.gitattributes

@@ -35,3 +35,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 test/decode_test.log filter=lfs diff=lfs merge=lfs -text
 test/hypo-132842.json filter=lfs diff=lfs merge=lfs -text
+test/vasr/audio/decode_test.log filter=lfs diff=lfs merge=lfs -text
+test/vasr/audio/hypo-132842.json filter=lfs diff=lfs merge=lfs -text
+test/vasr/audio_visual/decode_test.log filter=lfs diff=lfs merge=lfs -text
+test/vasr/audio_visual/hypo-132842.json filter=lfs diff=lfs merge=lfs -text

test/vasr/audio/cer.132842

@@ -0,0 +1,25 @@
+CER: 8.092690478125157
+err / num_ref_chars = 156363 / 1932151
+
+_name: null
+max_length: 300
+max_new_tokens: null
+min_length: 0
+min_new_tokens: null
+max_time: null
+do_sample: false
+num_beams: 20
+num_beam_groups: 1
+temperature: 1.0
+top_k: 50
+top_p: 1.0
+min_p: null
+typical_p: 1.0
+epsilon_cutoff: 0.0
+eta_cutoff: 0.0
+diversity_penalty: 0.0
+repetition_penalty: 1.0
+encoder_repetition_penalty: 1.0
+length_penalty: 1.0
+no_repeat_ngram_size: 0.0
+exponential_decay_length_penalty: null

test/vasr/audio/decode_test.log

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38be3c45a4fce70c3c4f03e8f7b80d994aa680f4efe1ff1aaa4d3f105f784545
+size 17238337

test/vasr/audio/hypo-132842.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0fc737bfeede00e79d13f762babf2a409d991ebaec6f62299c77832d018f091
+size 12000529

test/vasr/audio/wer.132842

@@ -0,0 +1,25 @@
+WER: 13.254824520514276
+err / num_ref_words = 62558 / 471964
+
+_name: null
+max_length: 300
+max_new_tokens: null
+min_length: 0
+min_new_tokens: null
+max_time: null
+do_sample: false
+num_beams: 20
+num_beam_groups: 1
+temperature: 1.0
+top_k: 50
+top_p: 1.0
+min_p: null
+typical_p: 1.0
+epsilon_cutoff: 0.0
+eta_cutoff: 0.0
+diversity_penalty: 0.0
+repetition_penalty: 1.0
+encoder_repetition_penalty: 1.0
+length_penalty: 1.0
+no_repeat_ngram_size: 0.0
+exponential_decay_length_penalty: null

test/vasr/audio_visual/cer.132842

@@ -0,0 +1,25 @@
+CER: 7.241048965634674
+err / num_ref_chars = 139908 / 1932151
+
+_name: null
+max_length: 300
+max_new_tokens: null
+min_length: 0
+min_new_tokens: null
+max_time: null
+do_sample: false
+num_beams: 20
+num_beam_groups: 1
+temperature: 1.0
+top_k: 50
+top_p: 1.0
+min_p: null
+typical_p: 1.0
+epsilon_cutoff: 0.0
+eta_cutoff: 0.0
+diversity_penalty: 0.0
+repetition_penalty: 1.0
+encoder_repetition_penalty: 1.0
+length_penalty: 1.0
+no_repeat_ngram_size: 0.0
+exponential_decay_length_penalty: null

test/vasr/audio_visual/decode_test.log

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1e88618adbd58943a01f40c7b61e0ef64647ea96a4c950f605ec27f282ccd29
+size 17267482

test/vasr/audio_visual/hypo-132842.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86a8444a245523bab1bb21611afa2d543f4b5e1eb63a389689e05c7977274774
+size 12026967

test/vasr/audio_visual/wer.132842

@@ -0,0 +1,25 @@
+WER: 12.029095439482672
+err / num_ref_words = 56773 / 471964
+
+_name: null
+max_length: 300
+max_new_tokens: null
+min_length: 0
+min_new_tokens: null
+max_time: null
+do_sample: false
+num_beams: 20
+num_beam_groups: 1
+temperature: 1.0
+top_k: 50
+top_p: 1.0
+min_p: null
+typical_p: 1.0
+epsilon_cutoff: 0.0
+eta_cutoff: 0.0
+diversity_penalty: 0.0
+repetition_penalty: 1.0
+encoder_repetition_penalty: 1.0
+length_penalty: 1.0
+no_repeat_ngram_size: 0.0
+exponential_decay_length_penalty: null