richardmfan commited on
Commit
273e128
·
verified ·
1 Parent(s): 3dd2c95

Upload sft_final

Browse files
chat_template.jinja ADDED
@@ -0,0 +1,134 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if tools %}
2
+ {{- "<|im_start|>system\n" }}
3
+ {%- if messages[0].role == 'system' and messages[0].content %}
4
+ {{- messages[0].content + '\n\n' }}
5
+ {%- endif %}
6
+ {{- "\n# Tools\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
7
+ {%- for tool in tools %}
8
+ {{- "\n" }}
9
+ {{- tool | tojson }}
10
+ {%- endfor %}
11
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
12
+ {%- else %}
13
+ {%- if messages[0].role == 'system' %}
14
+ {%- if messages[0].get('tools') %}
15
+ {%- if messages[0].content %}
16
+ {{- '<|im_start|>system\n' + messages[0].content + '\n\n' }}
17
+ {%- else %}
18
+ {{- '<|im_start|>system\n\n' }}
19
+ {%- endif %}
20
+ {{- "# Tools\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
21
+ {%- for tool in messages[0]['tools'] %}
22
+ {{- "\n" }}
23
+ {{- tool | tojson }}
24
+ {%- endfor %}
25
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
26
+ {%- else %}
27
+ {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
28
+ {%- endif %}
29
+ {%- endif %}
30
+ {%- endif %}
31
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
32
+ {%- for message in messages[::-1] %}
33
+ {%- set index = (messages|length - 1) - loop.index0 %}
34
+ {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
35
+ {%- set ns.multi_step_tool = false %}
36
+ {%- set ns.last_query_index = index %}
37
+ {%- endif %}
38
+ {%- endfor %}
39
+ {%- for message in messages %}
40
+ {%- if message.content is string %}
41
+ {%- set content = message.content %}
42
+ {%- else %}
43
+ {%- set content = '' %}
44
+ {%- endif %}
45
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
46
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
47
+ {%- elif message.role == "assistant" %}
48
+ {%- set thinking_content = '' %}
49
+ {%- set think_tag = '' %}
50
+ {%- if message.think is defined and message.think is string %}
51
+ {%- set thinking_content = message.think %}
52
+ {%- set think_tag = 'think' %}
53
+ {%- elif message.think_fast is defined and message.think_fast is string %}
54
+ {%- set thinking_content = message.think_fast %}
55
+ {%- set think_tag = 'think_fast' %}
56
+ {%- elif message.think_faster is defined and message.think_faster is string %}
57
+ {%- set thinking_content = message.think_faster %}
58
+ {%- set think_tag = 'think_faster' %}
59
+ {%- else %}
60
+ {%- if '</think>' in content %}
61
+ {%- set thinking_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
62
+ {%- set content = content.split('</think>')[-1].lstrip('\n') %}
63
+ {%- set think_tag = 'think' %}
64
+ {%- elif '</think_fast>' in content %}
65
+ {%- set thinking_content = content.split('</think_fast>')[0].rstrip('\n').split('<think_fast>')[-1].lstrip('\n') %}
66
+ {%- set content = content.split('</think_fast>')[-1].lstrip('\n') %}
67
+ {%- set think_tag = 'think_fast' %}
68
+ {%- elif '</think_faster>' in content %}
69
+ {%- set thinking_content = content.split('</think_faster>')[0].rstrip('\n').split('<think_faster>')[-1].lstrip('\n') %}
70
+ {%- set content = content.split('</think_faster>')[-1].lstrip('\n') %}
71
+ {%- set think_tag = 'think_faster' %}
72
+ {%- else %}
73
+ {%- set thinking_content = '' %}
74
+ {%- set think_tag = 'think_faster' %}
75
+ {%- endif %}
76
+ {%- endif %}
77
+ {{- '<|im_start|>' + message.role }}
78
+ {% generation %}
79
+ {%- if loop.index0 > ns.last_query_index %}
80
+ {%- if loop.last and think_tag %}
81
+ {%- if thinking_content %}
82
+ {{- '<' + think_tag + '>\n' + thinking_content + '\n</' + think_tag + '>\n' + content.lstrip('\n') }}
83
+ {%- else %}
84
+ {{- '<' + think_tag + '>\n</' + think_tag + '>\n' + content.lstrip('\n') }}
85
+ {%- endif %}
86
+ {%- else %}
87
+ {{- content }}
88
+ {%- endif %}
89
+ {%- else %}
90
+ {{- content }}
91
+ {%- endif %}
92
+ {%- if message.tool_calls %}
93
+ {%- for tool_call in message.tool_calls %}
94
+ {%- if (loop.first and content) or (not loop.first) %}
95
+ {{- '\n' }}
96
+ {%- endif %}
97
+ {%- if tool_call.function %}
98
+ {%- set tool_call = tool_call.function %}
99
+ {%- endif %}
100
+ {{- '<tool_call>\n{"name": "' }}
101
+ {{- tool_call.name }}
102
+ {{- '", "arguments": ' }}
103
+ {%- if tool_call.arguments is string %}
104
+ {{- tool_call.arguments }}
105
+ {%- else %}
106
+ {{- tool_call.arguments | tojson }}
107
+ {%- endif %}
108
+ {{- '}\n</tool_call>' }}
109
+ {%- endfor %}
110
+ {%- endif %}
111
+ {{- '<|im_end|>' }}
112
+ {% endgeneration %}
113
+ {%- elif message.role == "tool" %}
114
+ {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
115
+ {{- '<|im_start|>tool\n' }}
116
+ {%- endif %}
117
+ {{- content }}
118
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
119
+ {{- '<|im_end|>\n' }}
120
+ {%- endif %}
121
+ {%- endif %}
122
+ {%- endfor %}
123
+ {%- if add_generation_prompt %}
124
+ {%- set effort = reasoning_effort | default('medium') %}
125
+ {%- if effort == 'high' %}
126
+ {{- '<|im_start|>assistant\n<think>\n' }}
127
+ {%- elif effort == 'medium' %}
128
+ {{- '<|im_start|>assistant\n<think_fast>\n' }}
129
+ {%- elif effort == 'low' %}
130
+ {{- '<|im_start|>assistant\n<think_faster>\n' }}
131
+ {%- else %}
132
+ {{- '<|im_start|>assistant\n<think_fast>\n' }}
133
+ {%- endif %}
134
+ {%- endif %}
pytorch_model-00001-of-00006.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:72f394f456a393ab10eb47780c6e92f322ab23b868159e76f685e6f90e707fcc
3
  size 49871101268
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:909b8615f928423d2074ea7bf4ea31003cad7181a2800909239b4b9d5c157f14
3
  size 49871101268
pytorch_model-00002-of-00006.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:66cfc92ddb686bc4c03e786fe7aa6d086d5531720cb9ea7ff104f1dc2459e1c3
3
  size 49795740133
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:60bae6b0ced8f2cfa83b1f0d5b56511adce096d46e20d5ce4c33f68770ab1ef6
3
  size 49795740133
pytorch_model-00003-of-00006.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:209862eb9a9bef511d26cd08201d20756142aa0360d0713b7b43bac3709b263d
3
  size 49460263104
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d54ce154d56d1114bea3d8e5784f3b523f619e08bce8d28fe62a75e4f7adf3a
3
  size 49460263104
pytorch_model-00004-of-00006.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3083fd94e107ad70020800c4c8483a5a0da0aaca2f9b31fc4125fa54ad2185bc
3
  size 49795740133
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:29faf1fe1f7247a9067b263f58c6c3bbb83cff62064703f6e4d1ce1955adf822
3
  size 49795740133
pytorch_model-00005-of-00006.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:98c7990289d455e717945162f1e8fa64b1dbdd5ef0fb122c7d0bb2cb36af3edc
3
  size 49460263104
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1cf75748f7cc7116c8872f1e447531b7741a3fd90ff0e8d2e6b7d6ba5d96eaa9
3
  size 49460263104
pytorch_model-00006-of-00006.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fd0856cc39d2b53be106cd92a1d07fba6ad34afb9e08330363f0f5d3ce8f771e
3
  size 41817930473
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b61cc9e8a11fdd665f4d941509d3b947fed05523ffbac79479570b90a4d7138
3
  size 41817930473
tokenizer.json CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f4a4ca9d0a58168cb02f13950d12275d1cb3bccb470a907328f2d2e495116c73
3
- size 20680501
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:60c35bb84df92c4a57a70525008eed410589dfda476a63d43147fbbe85a443e7
3
+ size 20680378
tokenizer_config.json CHANGED
@@ -193,84 +193,84 @@
193
  "special": true
194
  },
195
  "250022": {
196
- "content": "reserved_special_token_0",
197
  "lstrip": false,
198
  "normalized": false,
199
  "rstrip": false,
200
  "single_word": false,
201
- "special": true
202
  },
203
  "250023": {
204
- "content": "reserved_special_token_1",
205
  "lstrip": false,
206
  "normalized": false,
207
  "rstrip": false,
208
  "single_word": false,
209
- "special": true
210
  },
211
  "250024": {
212
- "content": "reserved_special_token_2",
213
  "lstrip": false,
214
  "normalized": false,
215
  "rstrip": false,
216
  "single_word": false,
217
- "special": true
218
  },
219
  "250025": {
220
- "content": "reserved_special_token_3",
221
  "lstrip": false,
222
  "normalized": false,
223
  "rstrip": false,
224
  "single_word": false,
225
- "special": true
226
  },
227
  "250026": {
228
- "content": "reserved_special_token_4",
229
  "lstrip": false,
230
  "normalized": false,
231
  "rstrip": false,
232
  "single_word": false,
233
- "special": true
234
  },
235
  "250027": {
236
- "content": "reserved_special_token_5",
237
  "lstrip": false,
238
  "normalized": false,
239
  "rstrip": false,
240
  "single_word": false,
241
- "special": true
242
  },
243
  "250028": {
244
- "content": "reserved_special_token_6",
245
  "lstrip": false,
246
  "normalized": false,
247
  "rstrip": false,
248
  "single_word": false,
249
- "special": true
250
  },
251
  "250029": {
252
- "content": "reserved_special_token_7",
253
  "lstrip": false,
254
  "normalized": false,
255
  "rstrip": false,
256
  "single_word": false,
257
- "special": true
258
  },
259
  "250030": {
260
- "content": "reserved_special_token_8",
261
  "lstrip": false,
262
  "normalized": false,
263
  "rstrip": false,
264
  "single_word": false,
265
- "special": true
266
  },
267
  "250031": {
268
- "content": "reserved_special_token_9",
269
  "lstrip": false,
270
  "normalized": false,
271
  "rstrip": false,
272
  "single_word": false,
273
- "special": true
274
  },
275
  "250032": {
276
  "content": "reserved_special_token_10",
 
193
  "special": true
194
  },
195
  "250022": {
196
+ "content": "<think>",
197
  "lstrip": false,
198
  "normalized": false,
199
  "rstrip": false,
200
  "single_word": false,
201
+ "special": false
202
  },
203
  "250023": {
204
+ "content": "</think>",
205
  "lstrip": false,
206
  "normalized": false,
207
  "rstrip": false,
208
  "single_word": false,
209
+ "special": false
210
  },
211
  "250024": {
212
+ "content": "<think_fast>",
213
  "lstrip": false,
214
  "normalized": false,
215
  "rstrip": false,
216
  "single_word": false,
217
+ "special": false
218
  },
219
  "250025": {
220
+ "content": "</think_fast>",
221
  "lstrip": false,
222
  "normalized": false,
223
  "rstrip": false,
224
  "single_word": false,
225
+ "special": false
226
  },
227
  "250026": {
228
+ "content": "<think_faster>",
229
  "lstrip": false,
230
  "normalized": false,
231
  "rstrip": false,
232
  "single_word": false,
233
+ "special": false
234
  },
235
  "250027": {
236
+ "content": "</think_faster>",
237
  "lstrip": false,
238
  "normalized": false,
239
  "rstrip": false,
240
  "single_word": false,
241
+ "special": false
242
  },
243
  "250028": {
244
+ "content": "<tool_call>",
245
  "lstrip": false,
246
  "normalized": false,
247
  "rstrip": false,
248
  "single_word": false,
249
+ "special": false
250
  },
251
  "250029": {
252
+ "content": "</tool_call>",
253
  "lstrip": false,
254
  "normalized": false,
255
  "rstrip": false,
256
  "single_word": false,
257
+ "special": false
258
  },
259
  "250030": {
260
+ "content": "<tools>",
261
  "lstrip": false,
262
  "normalized": false,
263
  "rstrip": false,
264
  "single_word": false,
265
+ "special": false
266
  },
267
  "250031": {
268
+ "content": "</tools>",
269
  "lstrip": false,
270
  "normalized": false,
271
  "rstrip": false,
272
  "single_word": false,
273
+ "special": false
274
  },
275
  "250032": {
276
  "content": "reserved_special_token_10",