Commit
·
a2d26f2
1
Parent(s):
29d4128
Update README.md
Browse files
README.md
CHANGED
|
@@ -25,6 +25,8 @@ Regarding [benchmark scores](https://huggingface.co/spaces/JosephusCheung/Goodha
|
|
| 25 |
|
| 26 |
Disclaimer: Please note that the model was trained on unfiltered internet data. Since we do not have the capacity to vet all of it, there may be a substantial amount of objectionable content, pornography, violence, and offensive language present that we are unable to remove. Therefore, you will still need to complete your own checks on the model's safety and filter keywords in the output. Due to computational resource constraints, we are presently unable to implement RLHF for the model's ethics and safety, nor training on SFT samples that refuse to answer certain questions for restrictive fine-tuning.
|
| 27 |
|
|
|
|
|
|
|
| 28 |
# 迷你G
|
| 29 |
|
| 30 |
一个在超过1.2亿条数据合成数据集上训练的模型,这些数据集是通过应用具有大上下文窗口的最先进语言模型生成的,并结合了类似于检索增强生成和知识图谱集成的方法,数据合成是在一个由200亿个标记组成的预训练语料库中提取的聚类内进行的,随后由模型本身进行验证。
|
|
@@ -41,4 +43,6 @@ Disclaimer: Please note that the model was trained on unfiltered internet data.
|
|
| 41 |
|
| 42 |
关于[基准测试分数](https://huggingface.co/spaces/JosephusCheung/Goodharts-Law-on-Benchmarks-a-Page-for-miniG):一般来说,你不应该太过在意这些分数,因为人们总是可以专门训练以取得好成绩。我们主要将它们作为一个冒烟测试,一种快速检查,确保没有发生重大回退。事实上,如果你真的去阅读这些基准测试问题本身,你常常会发现自己会忍不住笑出声来,因为它们是多么无聊、低质量,甚至荒谬可笑。
|
| 43 |
|
| 44 |
-
免责声明:请注意,该模型是在未经过滤的互联网数据上训练的。由于我们无法对所有数据进行筛选,仍有可能存在大量不适当的内容——包括从露骨的材料到暴力和攻击性语言的内容——我们无法移除。因此,您必须自行对模型进行安全检查,并在输出中实施关键词过滤。由于计算资源的限制,我们目前无法为伦理和安全考虑进行人类反馈的强化学习(RLHF),也不能对SFT样本进行限制性微调,以限制模型回答某些问题的能力。
|
|
|
|
|
|
|
|
|
| 25 |
|
| 26 |
Disclaimer: Please note that the model was trained on unfiltered internet data. Since we do not have the capacity to vet all of it, there may be a substantial amount of objectionable content, pornography, violence, and offensive language present that we are unable to remove. Therefore, you will still need to complete your own checks on the model's safety and filter keywords in the output. Due to computational resource constraints, we are presently unable to implement RLHF for the model's ethics and safety, nor training on SFT samples that refuse to answer certain questions for restrictive fine-tuning.
|
| 27 |
|
| 28 |
+
Seeking Unconditional Sponsorship: We are actively training larger parameter models and scaling up data synthesis, and are seeking substantial compute resources and generous unconditional grants. While this is for the purpose of commercial exploration and technology selection, we are currently under no immediate pressure to generate profit and remain committed to sharing more with the open-source community.
|
| 29 |
+
|
| 30 |
# 迷你G
|
| 31 |
|
| 32 |
一个在超过1.2亿条数据合成数据集上训练的模型,这些数据集是通过应用具有大上下文窗口的最先进语言模型生成的,并结合了类似于检索增强生成和知识图谱集成的方法,数据合成是在一个由200亿个标记组成的预训练语料库中提取的聚类内进行的,随后由模型本身进行验证。
|
|
|
|
| 43 |
|
| 44 |
关于[基准测试分数](https://huggingface.co/spaces/JosephusCheung/Goodharts-Law-on-Benchmarks-a-Page-for-miniG):一般来说,你不应该太过在意这些分数,因为人们总是可以专门训练以取得好成绩。我们主要将它们作为一个冒烟测试,一种快速检查,确保没有发生重大回退。事实上,如果你真的去阅读这些基准测试问题本身,你常常会发现自己会忍不住笑出声来,因为它们是多么无聊、低质量,甚至荒谬可笑。
|
| 45 |
|
| 46 |
+
免责声明:请注意,该模型是在未经过滤的互联网数据上训练的。由于我们无法对所有数据进行筛选,仍有可能存在大量不适当的内容——包括从露骨的材料到暴力和攻击性语言的内容——我们无法移除。因此,您必须自行对模型进行安全检查,并在输出中实施关键词过滤。由于计算资源的限制,我们目前无法为伦理和安全考虑进行人类反馈的强化学习(RLHF),也不能对SFT样本进行限制性微调,以限制模型回答某些问题的能力。
|
| 47 |
+
|
| 48 |
+
寻求无条件赞助:我们正在积极训练更大参数的模型并扩大数据合成规模,同时寻求大量的计算资源和慷慨的无条件资助。尽管这是为了商业探索和技术选择的目的,但我们目前并没有立即产生利润的压力,并且仍然致力于与开源社区分享更多成果。
|