Upload tokenizer

2022-09-27 09:55:50 +00:00 · 2022-09-27 09:55:50 +00:00 · 8e72aa6038
parent a08dc56d11
commit 8e72aa6038
2 changed files with 9 additions and 2 deletions
--- a/special_tokens_map.json
+++ b/special_tokens_map.json
@ -108,5 +108,12 @@
    "<|notimestamps|>"
  ],
  "bos_token": "<|endoftext|>",
-  "eos_token": "<|endoftext|>"
+  "eos_token": "<|endoftext|>",
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
 }
--- a/tokenizer_config.json
+++ b/tokenizer_config.json
@ -19,7 +19,7 @@
  },
  "errors": "replace",
  "model_max_length": 1024,
-  "name_or_path": "/home/arthur_huggingface_co/whisper/_whisper/multi-hf",
+  "name_or_path": "openai/whisper-tiny",
  "pad_token": null,
  "processor_class": "WhisperProcessor",
  "special_tokens_map_file": null,