issues
search
InternLM
/
InternEvo
InternEvo is an open-sourced lightweight training framework aims to support model pre-training without the need for extensive dependencies.
https://internevo.readthedocs.io/zh-cn/latest/?badge=latest
Apache License 2.0
285
stars
47
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
[QA] loong train 支持packed_sample_into_one=false吗
#346
Lzhang-hub
opened
2 days ago
0
feat(moe): support group mlp for moe
#345
blankde
opened
5 days ago
0
feat(dataloader): refine implementation of mocked and megatron dataloader
#344
zigzagcai
opened
6 days ago
0
feat(zero bubble): update zbh1
#343
li126com
opened
6 days ago
0
[Bug] There will be timeout in some cases.
#342
kkscilife
closed
1 day ago
1
fix inject model and add multimodal dataloader
#341
sallyjunjun
closed
4 days ago
0
fix(enable_qkv_fusion): minor fix for qkv fusion
#340
zigzagcai
closed
1 week ago
0
fix dispatch model
#339
sallyjunjun
closed
1 week ago
0
fix(enable_qkv_fusion): refine wqkv fusion
#338
zigzagcai
closed
1 week ago
0
fix wqkv fusion
#337
zigzagcai
closed
1 week ago
0
fix wqkv fusion
#336
zigzagcai
closed
1 week ago
0
fix wqkv dim when enable qkv fusion
#335
sallyjunjun
closed
1 week ago
0
fix(pipeline): fix zero bubble pipeline parallelism
#334
li126com
closed
1 week ago
0
Feat(adam): support apex FusedAdam
#333
li126com
closed
1 week ago
0
feat(moe): add moe async param handler
#332
blankde
opened
2 weeks ago
0
feat(usability): Refine model inject helper to support huggingface models
#331
zigzagcai
closed
1 week ago
0
remove isp memory pool
#330
mwiacx
closed
2 weeks ago
0
update test loss
#329
li126com
opened
2 weeks ago
0
fix(isp): fix unnecessary module gather for isp
#328
blankde
closed
1 week ago
2
add qwen2moe and mixtral
#327
sallyjunjun
closed
1 week ago
1
feat(model: impl gpt 567 b
#326
blankde
opened
2 weeks ago
0
[Feature] MoE模型里稠密层和专家层zero和并行的解耦
#325
sunpengsdu
opened
2 weeks ago
0
[Feature] 不使用memory pool
#324
sunpengsdu
opened
2 weeks ago
1
feat(dataloader): Implement megatron dataloader and mocked dataloader
#323
zigzagcai
closed
2 weeks ago
1
feat(moe): support moe isp and no tp
#322
blankde
closed
2 weeks ago
0
feat(moe): support moe no tp
#321
blankde
closed
2 weeks ago
0
feat(moe): support dropless layer
#320
blankde
closed
2 weeks ago
3
fix(ci): fix weekly ci
#319
zigzagcai
closed
2 weeks ago
1
[Bug] There is an error in training : built-in model should inherited from BaseModel
#318
kkscilife
closed
2 weeks ago
1
fix(cross_entropy.py): replace the fa loss with apex loss
#317
yingtongxiong
closed
3 weeks ago
0
fix(shard.py): fix isp unpack data indexes err in rotary emb
#316
huangting4201
closed
3 weeks ago
0
add vacab parallel embedding
#315
mwiacx
closed
2 weeks ago
1
fix(ci): fix error in train_CI
#314
zigzagcai
closed
3 weeks ago
0
fix(model): fix bugs of batch generation & support min_new_tokens for inference
#313
x54-729
closed
3 weeks ago
0
Add new models
#312
sallyjunjun
closed
2 weeks ago
0
fix(embedding): fix incorrect computing of indexes in _update_cos_sin_cache
#311
li126com
closed
3 weeks ago
0
improve documentation
#310
sallyjunjun
closed
4 weeks ago
0
fix(910B): fix bugs in 910B for varlen and fixlen FA
#309
li126com
closed
1 month ago
2
fix(isp): fix dist-attn infer
#308
KimmiShi
closed
1 month ago
1
[Bug] 910B已知BUG和解决情况
#307
li126com
closed
3 weeks ago
0
[Feature] 优化ce_loss计算
#306
zigzagcai
closed
3 weeks ago
0
add data flow doc
#305
sallyjunjun
closed
1 month ago
0
feat(usability): Attempt for easier usability
#304
zigzagcai
closed
3 weeks ago
1
Attempt for easier usability
#303
zigzagcai
closed
1 month ago
0
[Bug] Import Error: Import "deeplink_ext.internlm_ops" could not be resolved
#302
kkscilife
closed
1 month ago
1
support pip install on npu environment
#301
sallyjunjun
closed
1 month ago
0
[QA] check import system var at the start of training
#300
sunpengsdu
opened
1 month ago
0
Zmz/qwen2
#299
hellozmz
closed
1 week ago
0
[Bug] 昇腾910安装internLM环境时报错需要nvcc
#298
tungsten106
closed
3 weeks ago
2
fix(launch): remove use_paked_data=use_flash_atten assert
#297
yingtongxiong
closed
3 weeks ago
0
Next