From 5764932c342e5fdfe7747f7183c600578fa72fa2 Mon Sep 17 00:00:00 2001 From: DrownFish19 Date: Mon, 9 Dec 2024 02:30:39 +0000 Subject: [PATCH 1/2] update docs/llm --- docs/llm/alignment/rm/flashmask/README.md | 1 + docs/llm/docs/predict/speculative_decoding.md | 1 + docs/llm/intel_hpu/llama/README.md | 1 + docs/llm/metax/llama/README.md | 1 + 4 files changed, 4 insertions(+) create mode 120000 docs/llm/alignment/rm/flashmask/README.md create mode 120000 docs/llm/docs/predict/speculative_decoding.md create mode 120000 docs/llm/intel_hpu/llama/README.md create mode 120000 docs/llm/metax/llama/README.md diff --git a/docs/llm/alignment/rm/flashmask/README.md b/docs/llm/alignment/rm/flashmask/README.md new file mode 120000 index 000000000000..f0bebc3d97ef --- /dev/null +++ b/docs/llm/alignment/rm/flashmask/README.md @@ -0,0 +1 @@ +../../../../../llm/alignment/rm/flashmask/README.md \ No newline at end of file diff --git a/docs/llm/docs/predict/speculative_decoding.md b/docs/llm/docs/predict/speculative_decoding.md new file mode 120000 index 000000000000..66f2ba69fa43 --- /dev/null +++ b/docs/llm/docs/predict/speculative_decoding.md @@ -0,0 +1 @@ +../../../../llm/docs/predict/speculative_decoding.md \ No newline at end of file diff --git a/docs/llm/intel_hpu/llama/README.md b/docs/llm/intel_hpu/llama/README.md new file mode 120000 index 000000000000..b93295220cf6 --- /dev/null +++ b/docs/llm/intel_hpu/llama/README.md @@ -0,0 +1 @@ +../../../../llm/intel_hpu/llama/README.md \ No newline at end of file diff --git a/docs/llm/metax/llama/README.md b/docs/llm/metax/llama/README.md new file mode 120000 index 000000000000..4ebc2c176df3 --- /dev/null +++ b/docs/llm/metax/llama/README.md @@ -0,0 +1 @@ +../../../../llm/metax/llama/README.md \ No newline at end of file From f348af6ab283c3816b96213b03114d270b92e1ea Mon Sep 17 00:00:00 2001 From: DrownFish19 Date: Mon, 9 Dec 2024 08:43:18 +0000 Subject: [PATCH 2/2] fix --- docs/slm/examples/benchmark/wiki_lambada/README.md | 1 + llm/README.md | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-) create mode 120000 docs/slm/examples/benchmark/wiki_lambada/README.md diff --git a/docs/slm/examples/benchmark/wiki_lambada/README.md b/docs/slm/examples/benchmark/wiki_lambada/README.md new file mode 120000 index 000000000000..b75d7740f65d --- /dev/null +++ b/docs/slm/examples/benchmark/wiki_lambada/README.md @@ -0,0 +1 @@ +../../../../../slm/examples/benchmark/wiki_lambada/README.md \ No newline at end of file diff --git a/llm/README.md b/llm/README.md index 7c297f752aea..90b34d5be2eb 100644 --- a/llm/README.md +++ b/llm/README.md @@ -50,7 +50,7 @@ PaddleNLP 将飞桨4D 并行策略加入到 Trainer API 中， 用户只需修 -我们在此处提供了更详细的[预训练数据制作](./tools/preprocess)，[Pretrain 和自定义数据集](https://paddlenlp.readthedocs.io/zh/latest/llm/pretraining/dataset.html)，[分布式策略支持情况](https://paddlenlp.readthedocs.io/zh/latest/llm/pretraining/index.html#model-capability)，[性能测试报告文档](https://paddlenlp.readthedocs.io/zh/latest/llm/pretraining/index.html#model-performance)，参见: [大模型预训练介绍](https://paddlenlp.readthedocs.io/zh/latest/llm/pretraining/index.html), [大模型权重列表](https://paddlenlp.readthedocs.io/zh/latest/llm/pretraining/index.html#model-weight)。 +我们在此处提供了更详细的[预训练数据制作](./tools/preprocess)，[Pretrain 和自定义数据集](https://paddlenlp.readthedocs.io/zh/latest/llm/dataset.html)，[分布式策略支持情况](https://paddlenlp.readthedocs.io/zh/latest/llm/docs/pretrain.html#model-capability)，[性能测试报告文档](https://paddlenlp.readthedocs.io/zh/latest/llm/docs/pretrain.html#model-performance)，参见: [大模型预训练介绍](https://paddlenlp.readthedocs.io/zh/latest/llm/docs/pretrain.html), [大模型权重列表](https://paddlenlp.readthedocs.io/zh/latest/llm/docs/pretrain.html#model-weight)。 此项目支持了 LLaMA、GPT-3、BaiChuan、Qwen 和 Mixtral 等大模型的预训练。用户切换配置 config 文件，即可一键运行。