diff --git a/README.md b/README.md index f5becce733cab1b8872b37b9230cb452dfaab240..95ad943650d6d9a0e5f932c829b964004b161002 100644 --- a/README.md +++ b/README.md @@ -429,9 +429,9 @@ DeepSparkHub甄选上百个应用算法和模型，覆盖AI和通用计算各领 -------- | ------ | ---- [ChatGLM-6B](nlp/llm/chatglm-6b/deepspeed/README.md) | PyTorch (DeepSpeed) | ADGEN & chatglm-6b [LLaMA-7B](nlp/llm/llama-7b/colossalai/README.md) | PyTorch (Colossal-AI) | llama-7b-hf -[LLaMA2-7B](nlp/llm/llama2-7b/megatron-deepspeed/README.md) | PyTorch (Megatron-DeepSpeed) | Bookcorpus -[LLaMA2-7B Reward Model Finetuning](nlp/llm/llama2-7b_reward_sft/deepspeed/README.md) | PyTorch (DeepSpeed) | Dahoas/rm-static -[LLaMA2-7B SFT](nlp/llm/llama2-7b_sft/megatron-deepspeed/README.md) | PyTorch (Megatron-DeepSpeed) | gpt_small-117M +[Llama-2-7B](nlp/llm/llama2-7b/megatron-deepspeed/README.md) | PyTorch (Megatron-DeepSpeed) | Bookcorpus +[Llama-2-7B Reward Model Finetuning](nlp/llm/llama2-7b_reward_sft/deepspeed/README.md) | PyTorch (DeepSpeed) | Dahoas/rm-static +[Llama-2-7B SFT](nlp/llm/llama2-7b_sft/megatron-deepspeed/README.md) | PyTorch (Megatron-DeepSpeed) | gpt_small-117M #### Text Correction diff --git a/RELEASE.md b/RELEASE.md index 66404c30acba3d952c88fe6b19a18c6544975b15..9a1b0fd1dd3be3ffb2d5a42f53ebfbbb8b7bc928 100644 --- a/RELEASE.md +++ b/RELEASE.md @@ -28,8 +28,8 @@ * 新增了基于分布式训练框架的大语言模型训练示例。 - * Megatron-DeepSpeed LLaMA2-7B SFT - * DeepSpeed LLaMa2-7B Reward Model Finetuning + * Megatron-DeepSpeed Llama-2-7B SFT + * DeepSpeed Llama-2-7B Reward Model Finetuning ### 问题修复 diff --git a/nlp/llm/llama2-7b/megatron-deepspeed/README.md b/nlp/llm/llama2-7b/megatron-deepspeed/README.md index 60718f5bd17743ac1442a91ae66861f8a61a0288..1ae436c935f29e24a865d4e13a4b630eba338475 100644 --- a/nlp/llm/llama2-7b/megatron-deepspeed/README.md +++ b/nlp/llm/llama2-7b/megatron-deepspeed/README.md @@ -1,8 +1,8 @@ -# Megatron-DeepSpeed LLaMA2-7B +# Megatron-DeepSpeed Llama-2-7B ## Model description -LLaMA2 is a large language model released by Meta in 2023, with parameters ranging from 7B to 70B. Compared to LLaMA, the training corpus of LLaMA2 is 40% longer, and the context length has been upgraded from 2048 to 4096, allowing for understanding and generating longer texts. +Llama 2 is a large language model released by Meta in 2023, with parameters ranging from 7B to 70B. Compared to LLaMA, the training corpus of Llama 2 is 40% longer, and the context length has been upgraded from 2048 to 4096, allowing for understanding and generating longer texts. ## Step 1: Installation diff --git a/nlp/llm/llama2-7b_reward_sft/deepspeed/README.md b/nlp/llm/llama2-7b_reward_sft/deepspeed/README.md index f72a66f3bb7f8bfab210fbc591df7ff6fbe73c50..e1f1d66ed97419753e00cd9ad2c120a21f734b81 100644 --- a/nlp/llm/llama2-7b_reward_sft/deepspeed/README.md +++ b/nlp/llm/llama2-7b_reward_sft/deepspeed/README.md @@ -1,4 +1,4 @@ -# LLaMa2 7B Reward Model Finetuning +# DeepSpeed Llama-2-7B Reward Model Finetuning ## Model description LLaMA2 is a large language model released by Meta in 2023, with parameters ranging from 7B to 70B. Compared to LLaMA, the training corpus of LLaMA2 is 40% longer, and the context length has been upgraded from 2048 to 4096, allowing for understanding and generating longer texts. diff --git a/nlp/llm/llama2-7b_sft/megatron-deepspeed/README.md b/nlp/llm/llama2-7b_sft/megatron-deepspeed/README.md index 62a52f657dbfcec36c23bfa6231ef6cf5e236270..17ec826ea85bd8ee39a6a19394fe0f158425cf4c 100644 --- a/nlp/llm/llama2-7b_sft/megatron-deepspeed/README.md +++ b/nlp/llm/llama2-7b_sft/megatron-deepspeed/README.md @@ -1,8 +1,8 @@ -# Megatron-DeepSpeed LLaMA2-7B SFT +# Megatron-DeepSpeed Llama-2-7B SFT ## Model description -LLaMA2 is a large language model released by Meta in 2023, with parameters ranging from 7B to 70B. Compared to LLaMA, the training corpus of LLaMA2 is 40% longer, and the context length has been upgraded from 2048 to 4096, allowing for understanding and generating longer texts. +Llama 2 is a large language model released by Meta in 2023, with parameters ranging from 7B to 70B. Compared to LLaMA, the training corpus of Llama 2 is 40% longer, and the context length has been upgraded from 2048 to 4096, allowing for understanding and generating longer texts. ## Step 1: Installation