diff --git a/docs/models/evaluation.md b/docs/models/evaluation.md
index e1dfaee3563f927180f3c04e2ad528bc0919e78c..f1c186c5f81f76150a3ffc506533c483641e7a5d 100644
--- a/docs/models/evaluation.md
+++ b/docs/models/evaluation.md
@@ -51,7 +51,7 @@ MindSpeed-LLM 已支持的大模型评估数据统计如下：
 | Yi1.5-9B             | MMLU   | 69.2%     | [69.5%](https://huggingface.co/01-ai/Yi-1.5-9B/tree/main)             | Yi1.5-34B        | MMLU   | 76.9%     | [77.1%](https://huggingface.co/01-ai/Yi-1.5-34B/tree/main)                         | 
 | CodeQWen2.5-7B       | Human. | 66.5%     | [61.6%](https://modelscope.cn/models/Qwen/Qwen2.5-Coder-7B)           | Qwen2.5-Math-7B  |MMLU-STEM| 65.1%    | [67.8%](https://github.com/QwenLM/Qwen2.5-Math/tree/main/)                         |
 | Qwen2.5-Math-72B     |MMLU-STEM| 83.7%    | [82.8%](https://github.com/QwenLM/Qwen2.5-Math/tree/main/)            | MiniCPM3-4B | MMLU | 63.7% | 64.6% |
-| Phi3.5-Mini          | MMLU   | 64.39%    | 64.34%                                                                |
+| Phi3.5-Mini          | MMLU   | 64.39%    | 64.34%                                                                | InternLM2-base-7b | MMLU | 63.1% | [64.0%](https://link.zhihu.com/?target=https%3A//arxiv.org/pdf/2403.17297.pdf)         |
 
 具体的权重转换功能命令介绍见 [MindSpeed-LLM 大模型使用指南分布式评估](../USER_GUIDE.md/#大模型分布式评估)章节.
 
diff --git a/docs/models/pretrain.md b/docs/models/pretrain.md
index 6775196d821dcb916b1caeca96689a0b6a2d6bd5..e43d7e53e3780937c1d3d597ac205f6ab559d664 100644
--- a/docs/models/pretrain.md
+++ b/docs/models/pretrain.md
@@ -174,6 +174,15 @@
       <td>【GTS】</td>
       <td>【Test】</td>
     </tr>
+    <tr>
+      <td rowspan="1"> <a href="https://huggingface.co/internlm">InternLM2-base</a> </td>
+      <td rowspan="1"> <a href="https://huggingface.co/internlm/internlm2-base-7b/tree/main">7B</a> </td>
+      <td> 4K </td>
+      <th> Mcore </th>
+      <td> 1x8 </td>
+      <td>【昇腾】</td>
+      <td>【Test】</td>
+    </tr>
     <tr>
       <td rowspan="3"> <a href="https://huggingface.co/internlm">InternLM2.5</a> </td>
       <td><a href="https://huggingface.co/internlm/internlm2_5-1_8b/tree/main">1.8B</a></td>
diff --git a/examples/mcore/internlm2/evaluate_internlm2_base_7b_ptd.sh b/examples/mcore/internlm2/evaluate_internlm2_base_7b_ptd.sh
new file mode 100644
index 0000000000000000000000000000000000000000..a98497199691304cf40e8b9f01163da63bc63dd0
--- /dev/null
+++ b/examples/mcore/internlm2/evaluate_internlm2_base_7b_ptd.sh
@@ -0,0 +1,61 @@
+#!/bin/bash
+export CUDA_DEVICE_MAX_CONNECTIONS=1
+
+# distributed config
+NPUS_PER_NODE=1
+MASTER_ADDR=localhost
+MASTER_PORT=6001
+NNODES=1
+NODE_RANK=0
+
+TOKENIZER_PATH="your tokenizer directory path"
+CHECKPOINT="your converted model checkpoint directory path"
+DATA_PATH="./data/mmlu/test"
+TASK="mmlu"
+
+DISTRIBUTED_ARGS="
+    --nproc_per_node $NPUS_PER_NODE \
+    --nnodes $NNODES \
+    --node_rank $NODE_RANK \
+    --master_addr $MASTER_ADDR \
+    --master_port $MASTER_PORT"
+
+torchrun ${DISTRIBUTED_ARGS} evaluation.py   \
+       --no-chat-template \
+       --task-data-path ${DATA_PATH} \
+       --task ${TASK} \
+       --use-mcore-models \
+       --tensor-model-parallel-size 1  \
+       --pipeline-model-parallel-size 1  \
+       --num-layers 32  \
+       --hidden-size 4096 \
+       --ffn-hidden-size 14336 \
+       --group-query-attention \
+       --num-query-groups 8 \
+       --position-embedding-type rope \
+       --norm-epsilon 1e-5 \
+       --rotary-base 1000000 \
+       --seq-length 32768 \
+       --max-new-tokens 1 \
+       --micro-batch-size 1  \
+       --num-attention-heads 32  \
+       --max-position-embeddings 32768 \
+       --padded-vocab-size 92544 \
+       --attention-dropout 0.0 \
+       --hidden-dropout 0.0 \
+       --swiglu \
+       --load ${CHECKPOINT} \
+       --disable-bias-linear \
+       --tokenizer-type PretrainedFromHF  \
+       --tokenizer-name-or-path ${TOKENIZER_PATH} \
+       --tokenizer-not-use-fast \
+       --normalization RMSNorm \
+       --exit-on-missing-checkpoint \
+       --no-load-rng \
+       --no-load-optim \
+       --untie-embeddings-and-output-weights \
+       --no-masked-softmax-fusion \
+       --no-gradient-accumulation-fusion \
+       --make-vocab-size-divisible-by 1 \
+       --seed 42 \
+       | tee logs/evaluate_internlm2_base_7b_mcore_${TASK}.log
\ No newline at end of file
diff --git a/examples/mcore/internlm2/generate_internlm2_base_7b_ptd.sh b/examples/mcore/internlm2/generate_internlm2_base_7b_ptd.sh
new file mode 100644
index 0000000000000000000000000000000000000000..c377df89ca3d969ae962173260f8bfc1dcaa9312
--- /dev/null
+++ b/examples/mcore/internlm2/generate_internlm2_base_7b_ptd.sh
@@ -0,0 +1,60 @@
+      
+#!/bin/bash
+
+# The number of parameters is not aligned
+export LD_LIBRARY_PATH=/usr/local/lib:/usr/local/lib:/root/miniconda3/lib:$LD_LIBRARY_PATH
+export HCCL_CONNECT_TIMEOUT=1200
+export COMBINED_ENABLE=1
+export CUDA_DEVICE_MAX_CONNECTIONS=1
+
+# please fill these path configurations
+CHECKPOINT="your model directory path"
+TOKENIZER_PATH="your tokenizer directory path"
+TOKENIZER_MODEL="your tokenizer.model file path"
+
+# Change for multinode config
+MASTER_ADDR=localhost
+MASTER_PORT=6001
+NNODES=1
+NODE_RANK=0
+NPUS_PER_NODE=1
+WORLD_SIZE=$(($NPUS_PER_NODE*$NNODES))
+
+DISTRIBUTED_ARGS="--nproc_per_node $NPUS_PER_NODE --nnodes $NNODES --node_rank $NODE_RANK --master_addr $MASTER_ADDR --master_port $MASTER_PORT"
+
+python -m torch.distributed.launch $DISTRIBUTED_ARGS inference.py \
+       --tensor-model-parallel-size 1  \
+       --pipeline-model-parallel-size 1  \
+       --num-layers 32 \
+       --hidden-size 4096  \
+       --ffn-hidden-size 14336 \
+       --position-embedding-type rope \
+       --use-rotary-position-embeddings \
+       --rotary-base 1000000 \
+       --group-query-attention \
+       --num-query-groups 8 \
+       --seq-length 8192 \
+       --max-new-tokens 256 \
+       --micro-batch-size 1 \
+       --num-attention-heads 32  \
+       --max-position-embeddings 32768 \
+       --padded-vocab-size 92544 \
+       --attention-dropout 0.0 \
+       --hidden-dropout 0.0 \
+       --swiglu \
+       --load "${CHECKPOINT}"  \
+       --tokenizer-type PretrainedFromHF  \
+       --tokenizer-name-or-path "${TOKENIZER_PATH}" \
+       --tokenizer-model "${TOKENIZER_MODEL}"  \
+       --tokenizer-not-use-fast \
+       --bf16 \
+       --normalization RMSNorm \
+       --norm-epsilon 1e-5 \
+       --untie-embeddings-and-output-weights \
+       --disable-bias-linear \
+       --attention-softmax-in-fp32 \
+       --exit-on-missing-checkpoint \
+       --seed 42 \
+       --make-vocab-size-divisible-by 1 \
+       --use-mcore-models \
+       | tee logs/generate_internlm2_base_7b.log
\ No newline at end of file
diff --git a/examples/mcore/internlm2/pretrain_internlm2_base_7b_ptd.sh b/examples/mcore/internlm2/pretrain_internlm2_base_7b_ptd.sh
new file mode 100644
index 0000000000000000000000000000000000000000..72882cd7b397334044e1f768cde2d76c9ad358f5
--- /dev/null
+++ b/examples/mcore/internlm2/pretrain_internlm2_base_7b_ptd.sh
@@ -0,0 +1,101 @@
+#!/bin/bash
+
+export CUDA_DEVICE_MAX_CONNECTIONS=1
+export HCCL_CONNECT_TIMEOUT=1200
+
+GPUS_PER_NODE=2
+MASTER_ADDR=localhost
+MASTER_PORT=6000
+NNODES=1
+NODE_RANK=0
+WORLD_SIZE=$(($GPUS_PER_NODE*$NNODES))
+
+CKPT_SAVE_DIR="your model save ckpt path"
+DATA_PATH="your data path"
+TOKENIZER_MODEL="your tokenizer path"
+CKPT_LOAD_DIR="your model ckpt path, the parameters TP and PP should be the same as the converted model"
+TP=1
+PP=2
+
+DISTRIBUTED_ARGS="
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --node_rank $NODE_RANK \
+    --master_addr $MASTER_ADDR \
+    --master_port $MASTER_PORT
+"
+
+GPT_ARGS="
+    --use-mcore-models \
+    --sequence-parallel \
+    --tensor-model-parallel-size ${TP} \
+    --pipeline-model-parallel-size ${PP} \
+    --num-layers 32 \
+    --hidden-size 4096 \
+    --ffn-hidden-size 14336 \
+    --num-attention-heads 32 \
+    --group-query-attention \
+    --num-query-groups 8 \
+    --tokenizer-type PretrainedFromHF \
+    --tokenizer-name-or-path ${TOKENIZER_MODEL} \
+    --seq-length 256 \
+    --max-position-embeddings 256 \
+    --micro-batch-size 1 \
+    --global-batch-size 1 \
+    --make-vocab-size-divisible-by 1 \
+    --padded-vocab-size 92544 \
+    --lr 1e-6 \
+    --lr-decay-style cosine \
+    --rotary-base 1000000 \
+    --train-iters 1 \
+    --untie-embeddings-and-output-weights \
+    --disable-bias-linear \
+    --init-method-std 0.01 \
+    --attention-dropout 0.0 \
+    --hidden-dropout 0.0 \
+    --min-lr 1.25e-7 \
+    --weight-decay 1e-1 \
+    --lr-warmup-fraction 0.01 \
+    --clip-grad 1.0 \
+    --adam-beta1 0.9 \
+    --adam-beta2 0.95 \
+    --initial-loss-scale 4096 \
+    --no-gradient-accumulation-fusion \
+    --use-flash-attn \
+    --use-fused-rotary-pos-emb \
+    --use-rotary-position-embeddings \
+    --use-fused-swiglu \
+    --use-fused-rmsnorm \
+    --position-embedding-type rope \
+    --normalization RMSNorm \
+    --norm-epsilon 1e-5 \
+    --swiglu \
+    --no-masked-softmax-fusion \
+    --attention-softmax-in-fp32 \
+    --no-load-optim \
+    --no-load-rng \
+    --seed 42 \
+    --bf16
+"
+    # --use-mc2 \
+
+DATA_ARGS="
+    --data-path $DATA_PATH \
+    --split 100,0,0
+"
+
+OUTPUT_ARGS="
+    --log-interval 1 \
+    --save-interval 2000 \
+    --eval-interval 2000 \
+    --eval-iters 0 \
+"
+
+torchrun $DISTRIBUTED_ARGS pretrain_gpt.py \
+    $GPT_ARGS \
+    $DATA_ARGS \
+    $OUTPUT_ARGS \
+    --distributed-backend nccl \
+    --load ${CKPT_LOAD_DIR} \
+    --save ${CKPT_SAVE_DIR} \
+    | tee logs/train_internlm2_base_7b_micro.log
\ No newline at end of file