Upload folder using huggingface_hub
Browse files- README.md +4 -0
- chat_template.jinja +5 -0
- special_tokens_map.json +2 -14
- tokenizer_config.json +2 -2
README.md
CHANGED
|
@@ -37,6 +37,10 @@ save_folder = "/tmp/yujiepan/baguettotron-tiny-random"
|
|
| 37 |
tokenizer = AutoTokenizer.from_pretrained(
|
| 38 |
source_model_id, trust_remote_code=True,
|
| 39 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
tokenizer.save_pretrained(save_folder)
|
| 41 |
|
| 42 |
config = AutoConfig.from_pretrained(
|
|
|
|
| 37 |
tokenizer = AutoTokenizer.from_pretrained(
|
| 38 |
source_model_id, trust_remote_code=True,
|
| 39 |
)
|
| 40 |
+
tokenizer.chat_template = "{% for m in messages %}<|im_start|>{{ m['role'] }}\n{{ m['content'] }}<|im_end|>\n{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n<think>\n{% endif %}"
|
| 41 |
+
tokenizer.eos_token = "<|im_end|>"
|
| 42 |
+
tokenizer.bos_token = "<|im_start|>"
|
| 43 |
+
tokenizer.stop_token = "<|im_end|>"
|
| 44 |
tokenizer.save_pretrained(save_folder)
|
| 45 |
|
| 46 |
config = AutoConfig.from_pretrained(
|
chat_template.jinja
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{% for m in messages %}<|im_start|>{{ m['role'] }}
|
| 2 |
+
{{ m['content'] }}<|im_end|>
|
| 3 |
+
{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant
|
| 4 |
+
<think>
|
| 5 |
+
{% endif %}
|
special_tokens_map.json
CHANGED
|
@@ -46,20 +46,8 @@
|
|
| 46 |
"⟨H≈1.7⟩",
|
| 47 |
"⟨H≈1.8⟩"
|
| 48 |
],
|
| 49 |
-
"bos_token":
|
| 50 |
-
|
| 51 |
-
"lstrip": false,
|
| 52 |
-
"normalized": false,
|
| 53 |
-
"rstrip": false,
|
| 54 |
-
"single_word": false
|
| 55 |
-
},
|
| 56 |
-
"eos_token": {
|
| 57 |
-
"content": "<|end_of_text|>",
|
| 58 |
-
"lstrip": false,
|
| 59 |
-
"normalized": false,
|
| 60 |
-
"rstrip": false,
|
| 61 |
-
"single_word": false
|
| 62 |
-
},
|
| 63 |
"pad_token": {
|
| 64 |
"content": "[PAD]",
|
| 65 |
"lstrip": false,
|
|
|
|
| 46 |
"⟨H≈1.7⟩",
|
| 47 |
"⟨H≈1.8⟩"
|
| 48 |
],
|
| 49 |
+
"bos_token": "<|im_start|>",
|
| 50 |
+
"eos_token": "<|im_end|>",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
"pad_token": {
|
| 52 |
"content": "[PAD]",
|
| 53 |
"lstrip": false,
|
tokenizer_config.json
CHANGED
|
@@ -440,9 +440,9 @@
|
|
| 440 |
"⟨H≈1.7⟩",
|
| 441 |
"⟨H≈1.8⟩"
|
| 442 |
],
|
| 443 |
-
"bos_token": "<|
|
| 444 |
"clean_up_tokenization_spaces": true,
|
| 445 |
-
"eos_token": "<|
|
| 446 |
"extra_special_tokens": {},
|
| 447 |
"model_max_length": 1000000000000000019884624838656,
|
| 448 |
"pad_token": "[PAD]",
|
|
|
|
| 440 |
"⟨H≈1.7⟩",
|
| 441 |
"⟨H≈1.8⟩"
|
| 442 |
],
|
| 443 |
+
"bos_token": "<|im_start|>",
|
| 444 |
"clean_up_tokenization_spaces": true,
|
| 445 |
+
"eos_token": "<|im_end|>",
|
| 446 |
"extra_special_tokens": {},
|
| 447 |
"model_max_length": 1000000000000000019884624838656,
|
| 448 |
"pad_token": "[PAD]",
|