能不能給個 ollama的下載
#1
by
woolkingx
- opened
- README.md +4 -4
- config.json +3 -3
- generation_config.json +1 -1
- model-00001-of-00002.safetensors +1 -1
- model-00002-of-00002.safetensors +2 -2
- model.safetensors.index.json +1 -2
- tokenizer_config.json +1 -1
README.md
CHANGED
@@ -24,7 +24,7 @@ library_name: transformers
|
|
24 |
因為簡體中文和繁體中文在語法和語義相似度高,往往只是差在文字本身看起來不一樣。另外,千問 (Qwen) 的模型中文的能力是很強的 (很可惜 3B 的 license 不是 apache)。我們利用[繁化姬](https://zhconvert.org) 將簡體中文所對應的 token 和 繁體中文的 token 做代換。更多關於 tokenizer 帶換掉哪些單詞,可以查看 [benchang1110/DeepSeek-R1-Distill-Qwen-1.5B-zhtw](https://huggingface.co/benchang1110/DeepSeek-R1-Distill-Qwen-1.5B-zhtw) 中的 convert.txt。
|
25 |
|
26 |
### SFT
|
27 |
-
|
28 |
|
29 |
### DPO
|
30 |
這個階段我們用 [zake7749/kyara-chinese-preference-rl-dpo-s0-30K](https://huggingface.co/datasets/zake7749/kyara-chinese-preference-rl-dpo-s0-30K) 這個資料集做 alignment,目標讓模型能夠條列式的輸出,並且能夠在輸出的時候保持一定的邏輯性。
|
@@ -130,7 +130,8 @@ ASSISTANT:在我的時代,即18世紀,音樂風格經歷了顯著的發展
|
|
130 |
|
131 |
## Bias, Risks, and Limitations
|
132 |
|
133 |
-
修改 system prompt
|
|
|
134 |
|
135 |
## How to Get Started with the Model
|
136 |
|
@@ -358,8 +359,7 @@ GPU Hours: A100*3h
|
|
358 |
|
359 |
|
360 |
## Summary
|
361 |
-
這個模型借用了 base model 強大的能力,和同級的模型相比在 TMLU 和 TMMLU+ 上有更好的表現,甚至可以和 8B
|
362 |
-
開發成本也比上述的模型低非常多。在日常對話和繁體中文上表現都相當優異。感謝提供 base model 和資料集的團隊。
|
363 |
|
364 |
## Citation
|
365 |
|
|
|
24 |
因為簡體中文和繁體中文在語法和語義相似度高,往往只是差在文字本身看起來不一樣。另外,千問 (Qwen) 的模型中文的能力是很強的 (很可惜 3B 的 license 不是 apache)。我們利用[繁化姬](https://zhconvert.org) 將簡體中文所對應的 token 和 繁體中文的 token 做代換。更多關於 tokenizer 帶換掉哪些單詞,可以查看 [benchang1110/DeepSeek-R1-Distill-Qwen-1.5B-zhtw](https://huggingface.co/benchang1110/DeepSeek-R1-Distill-Qwen-1.5B-zhtw) 中的 convert.txt。
|
25 |
|
26 |
### SFT
|
27 |
+
這個階段我們用 [lianghsun/tw-instruct-500k](https://huggingface.co/datasets/lianghsun/tw-instruct-500k) 這個資料集做微調,目標讓模型知道一點台灣的在地文化和一些台灣常用的慣用語,但為了保持模型原先強大的性能,我們用 LoRA 微調。
|
28 |
|
29 |
### DPO
|
30 |
這個階段我們用 [zake7749/kyara-chinese-preference-rl-dpo-s0-30K](https://huggingface.co/datasets/zake7749/kyara-chinese-preference-rl-dpo-s0-30K) 這個資料集做 alignment,目標讓模型能夠條列式的輸出,並且能夠在輸出的時候保持一定的邏輯性。
|
|
|
130 |
|
131 |
## Bias, Risks, and Limitations
|
132 |
|
133 |
+
修改 system prompt 會調整模型輸出的偏見和立場(~~這裡就不多提了,各位自己調調看~~)。
|
134 |
+
|
135 |
|
136 |
## How to Get Started with the Model
|
137 |
|
|
|
359 |
|
360 |
|
361 |
## Summary
|
362 |
+
這個模型借用了 base model 強大的能力,和同級的模型相比在 TMLU 和 TMMLU+ 上有更好的表現,甚至可以和 8B 的模型相比。開發成本也比上述的模型低非常多。撇除一些撇除一些小缺點(小機率出現簡體中文),在日常對話和對台灣的理解上表現都相當優異。感謝提供 base model 和資料集的團隊。
|
|
|
363 |
|
364 |
## Citation
|
365 |
|
config.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "./DPO/
|
3 |
"architectures": [
|
4 |
"Qwen2ForCausalLM"
|
5 |
],
|
@@ -20,9 +20,9 @@
|
|
20 |
"rope_scaling": null,
|
21 |
"rope_theta": 1000000.0,
|
22 |
"sliding_window": null,
|
23 |
-
"tie_word_embeddings":
|
24 |
"torch_dtype": "bfloat16",
|
25 |
-
"transformers_version": "4.
|
26 |
"use_cache": true,
|
27 |
"use_sliding_window": false,
|
28 |
"vocab_size": 151936
|
|
|
1 |
{
|
2 |
+
"_name_or_path": "./DPO/Qwen_3B_tw_dpo_phase2",
|
3 |
"architectures": [
|
4 |
"Qwen2ForCausalLM"
|
5 |
],
|
|
|
20 |
"rope_scaling": null,
|
21 |
"rope_theta": 1000000.0,
|
22 |
"sliding_window": null,
|
23 |
+
"tie_word_embeddings": true,
|
24 |
"torch_dtype": "bfloat16",
|
25 |
+
"transformers_version": "4.48.0",
|
26 |
"use_cache": true,
|
27 |
"use_sliding_window": false,
|
28 |
"vocab_size": 151936
|
generation_config.json
CHANGED
@@ -10,5 +10,5 @@
|
|
10 |
"temperature": 0.7,
|
11 |
"top_k": 20,
|
12 |
"top_p": 0.8,
|
13 |
-
"transformers_version": "4.
|
14 |
}
|
|
|
10 |
"temperature": 0.7,
|
11 |
"top_k": 20,
|
12 |
"top_p": 0.8,
|
13 |
+
"transformers_version": "4.48.0"
|
14 |
}
|
model-00001-of-00002.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4957560304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3f87325420b90d2446553839391755109367dcd961636cb9a1f8c271c23dff0f
|
3 |
size 4957560304
|
model-00002-of-00002.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6df34840f566355e74762a79b4512065f703c5737259083536acea92100308c2
|
3 |
+
size 1214366696
|
model.safetensors.index.json
CHANGED
@@ -1,9 +1,8 @@
|
|
1 |
{
|
2 |
"metadata": {
|
3 |
-
"total_size":
|
4 |
},
|
5 |
"weight_map": {
|
6 |
-
"lm_head.weight": "model-00002-of-00002.safetensors",
|
7 |
"model.embed_tokens.weight": "model-00001-of-00002.safetensors",
|
8 |
"model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
9 |
"model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
|
|
1 |
{
|
2 |
"metadata": {
|
3 |
+
"total_size": 6171877376
|
4 |
},
|
5 |
"weight_map": {
|
|
|
6 |
"model.embed_tokens.weight": "model-00001-of-00002.safetensors",
|
7 |
"model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
|
8 |
"model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
|
tokenizer_config.json
CHANGED
@@ -195,7 +195,7 @@
|
|
195 |
"<|video_pad|>"
|
196 |
],
|
197 |
"bos_token": null,
|
198 |
-
"chat_template": "{%- if tools %}\n {{- '<|im_start|>system\\n' }}\n {%- if messages[0]['role'] == 'system' %}\n {{- messages[0]['content'] }}\n {%- else %}\n {{- '你是 Qwen-Taiwan-3B,
|
199 |
"clean_up_tokenization_spaces": false,
|
200 |
"eos_token": "<|im_end|>",
|
201 |
"errors": "replace",
|
|
|
195 |
"<|video_pad|>"
|
196 |
],
|
197 |
"bos_token": null,
|
198 |
+
"chat_template": "{%- if tools %}\n {{- '<|im_start|>system\\n' }}\n {%- if messages[0]['role'] == 'system' %}\n {{- messages[0]['content'] }}\n {%- else %}\n {{- '你是 Qwen-Taiwan-3B, 來自台灣新竹市。你是一位樂於回答問題的助手。' }}\n {%- endif %}\n {{- \"\\n\\n# Tools\\n\\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>\" }}\n {%- for tool in tools %}\n {{- \"\\n\" }}\n {{- tool | tojson }}\n {%- endfor %}\n {{- \"\\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\" }}\n{%- else %}\n {%- if messages[0]['role'] == 'system' %}\n {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n {%- else %}\n {{- '<|im_start|>system\\n你是 Qwen-Taiwan-3B, 來自台灣新竹市。你是一位樂於回答問題的助手。<|im_end|>\\n' }}\n {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n {%- elif message.role == \"assistant\" %}\n {{- '<|im_start|>' + message.role }}\n {%- if message.content %}\n {{- '\\n' + message.content }}\n {%- endif %}\n {%- for tool_call in message.tool_calls %}\n {%- if tool_call.function is defined %}\n {%- set tool_call = tool_call.function %}\n {%- endif %}\n {{- '\\n<tool_call>\\n{\"name\": \"' }}\n {{- tool_call.name }}\n {{- '\", \"arguments\": ' }}\n {{- tool_call.arguments | tojson }}\n {{- '}\\n</tool_call>' }}\n {%- endfor %}\n {{- '<|im_end|>\\n' }}\n {%- elif message.role == \"tool\" %}\n {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n {{- '<|im_start|>user' }}\n {%- endif %}\n {{- '\\n<tool_response>\\n' }}\n {{- message.content }}\n {{- '\\n</tool_response>' }}\n {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n {{- '<|im_end|>\\n' }}\n {%- endif %}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
|
199 |
"clean_up_tokenization_spaces": false,
|
200 |
"eos_token": "<|im_end|>",
|
201 |
"errors": "replace",
|