Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / mosaicml/llm-foundry issues and pull requests

#1303 - Bumping mlflow version to include buffering

Pull Request - State: closed - Opened by JackZ-db 4 days ago

#1302 - Add Retries to run_query

Pull Request - State: closed - Opened by KuuCi 4 days ago - 1 comment

#1300 - Add `all` transforms to train script

Pull Request - State: closed - Opened by dakinggg 6 days ago

#1298 - Allow passing in lbl_process_group directly

Pull Request - State: closed - Opened by dakinggg 7 days ago

#1297 - Bump composer to 0.23.4

Pull Request - State: closed - Opened by mvpatel2000 7 days ago

#1296 - Fix grad accum typing

Pull Request - State: closed - Opened by dakinggg 7 days ago

#1295 - [Do Not Merge] Test patch

Pull Request - State: closed - Opened by mvpatel2000 8 days ago

#1294 - Bump min composer version to 0.23.3

Pull Request - State: closed - Opened by dakinggg 8 days ago

#1293 - Small refactor for update batch size

Pull Request - State: closed - Opened by dakinggg 8 days ago

#1292 - Removing logging exception through update run metadata

Pull Request - State: open - Opened by jjanezhang 9 days ago

#1291 - Unable to use self developed pre-trained model for finetuning in MosaicML

Issue - State: open - Opened by sauravgrd 9 days ago - 1 comment
Labels: question

#1290 - Extendability refactors

Pull Request - State: closed - Opened by dakinggg 9 days ago - 1 comment

#1289 - MPT training with ALiBi and Flash Attention 2

Issue - State: open - Opened by rickgit16 10 days ago - 3 comments
Labels: question

#1288 - Add TE to setup

Pull Request - State: closed - Opened by j316chuck 11 days ago

#1287 - Add missing dependency group

Pull Request - State: closed - Opened by dakinggg 14 days ago - 1 comment

#1286 - Fix backwards compatibility for ICL arg

Pull Request - State: closed - Opened by dakinggg 14 days ago

#1283 - Try to fix eval_output_logging_callback.py

Pull Request - State: open - Opened by sjawhar 15 days ago - 6 comments

#1280 - Fix TE HF checkpoint saving

Pull Request - State: closed - Opened by j316chuck 15 days ago - 1 comment

#1277 - Fix packing + streaming + resumption

Pull Request - State: closed - Opened by dakinggg 15 days ago

#1276 - Allow multiprocessing when preparing ICL dataset

Issue - State: open - Opened by sanjari-orb 16 days ago - 8 comments
Labels: enhancement

#1274 - Update Dockerfile

Pull Request - State: open - Opened by j316chuck 17 days ago

#1273 - Update Dockerfile with TE main

Pull Request - State: open - Opened by j316chuck 17 days ago

#1271 - Why is there a warmup in hf_generate.py?

Issue - State: open - Opened by palash04 17 days ago
Labels: question

#1270 - fix linting

Pull Request - State: closed - Opened by milocress 20 days ago

#1269 - Bump Composer to version 0.23.2

Pull Request - State: closed - Opened by dakinggg 21 days ago

#1268 - Revert "Bump Composer to 0.23.0 (#1259)"

Pull Request - State: closed - Opened by dakinggg 21 days ago

#1267 - Revert to older TE version

Pull Request - State: closed - Opened by mvpatel2000 21 days ago - 1 comment

#1266 - Revert "Update TE Dockerfile (#1265)"

Pull Request - State: closed - Opened by j316chuck 21 days ago

#1265 - Update TE Dockerfile

Pull Request - State: closed - Opened by j316chuck 21 days ago

#1264 - Fill in the middle

Issue - State: open - Opened by germanjke 21 days ago

#1263 - Fix typo in setup.py

Pull Request - State: closed - Opened by XiaohanZhangCMU 21 days ago

#1262 - Testing CI

Pull Request - State: closed - Opened by dakinggg 21 days ago

#1261 - How to continue pretrain LLM fp8 with hf_causal_lm

Issue - State: open - Opened by YixinSong-e 21 days ago
Labels: bug

#1260 - added systemMetricsMonitor callback

Pull Request - State: closed - Opened by JackZ-db 22 days ago - 1 comment

#1259 - Bump Composer to 0.23.0

Pull Request - State: closed - Opened by KuuCi 22 days ago - 1 comment

#1258 - Remove spurious warning

Pull Request - State: closed - Opened by dakinggg 22 days ago

#1257 - Fix MPT HF conversion

Pull Request - State: closed - Opened by dakinggg 23 days ago

#1256 - Add curriculum learning callback

Pull Request - State: open - Opened by b-chu 23 days ago - 3 comments

#1255 - Bump Version to 0.10.0.dev0

Pull Request - State: closed - Opened by KuuCi 23 days ago - 2 comments

#1254 - Adding more token encoding types

Pull Request - State: closed - Opened by snarayan21 23 days ago - 1 comment

#1253 - fix signal_file_path to avoid race condition

Pull Request - State: open - Opened by ofivite 23 days ago - 10 comments

#1252 - Add registry for ICL datasets

Pull Request - State: open - Opened by sanjari-orb 24 days ago - 5 comments

#1251 - Change TE docker image to enable te_shard_weight

Pull Request - State: closed - Opened by j316chuck 24 days ago

#1249 - Testing CI

Pull Request - State: closed - Opened by dakinggg 25 days ago

#1248 - Update CODEOWNERS

Pull Request - State: closed - Opened by dakinggg 25 days ago

#1247 - Add eval_drop_last flag to fix TE eval bug

Pull Request - State: open - Opened by j316chuck 25 days ago - 2 comments

#1244 - Fix the error message thrown from dataloader

Pull Request - State: open - Opened by shitaoli-db 29 days ago

#1243 - Add logging to convert_text_to_mds.py script

Pull Request - State: closed - Opened by irenedea 30 days ago

#1242 - could you give an elaborated steps about how to run llm-foundry on AMD mi250 devices

Issue - State: open - Opened by Alice1069 about 1 month ago - 1 comment
Labels: bug

#1241 - Make HF conversion automatically add missing imports

Pull Request - State: closed - Opened by dakinggg about 1 month ago

#1240 - Chunk file reads and tokenization for text to mds conversion

Pull Request - State: closed - Opened by irenedea about 1 month ago - 1 comment

#1239 - Make the exceptions serializable

Pull Request - State: closed - Opened by dakinggg about 1 month ago

#1238 - Add retries to downloads in convert_text_to_mds.py

Pull Request - State: closed - Opened by irenedea about 1 month ago - 1 comment

#1237 - [WIP] use_remote_uploader_v2

Pull Request - State: open - Opened by bigning about 1 month ago

#1236 - Configurable submesh

Pull Request - State: closed - Opened by dakinggg about 1 month ago

#1235 - Fix tuple typing

Pull Request - State: closed - Opened by dakinggg about 1 month ago

#1234 - Move MLFlow dataset outside of log_config

Pull Request - State: closed - Opened by KuuCi about 1 month ago - 1 comment

#1233 - Fix Mosaic Logger custom exception serialization

Pull Request - State: closed - Opened by milocress about 1 month ago

#1231 - LLaMA PRO training resume problem

Issue - State: open - Opened by germanjke about 1 month ago - 5 comments

#1230 - Fix attr error for attention_classes when using act ckpt

Pull Request - State: closed - Opened by cli99 about 1 month ago

#1229 - Modularize backbone class and block creation

Pull Request - State: closed - Opened by dakinggg about 1 month ago

#1228 - Quick patch to check that Dataset Keys contain non-None Values

Pull Request - State: closed - Opened by KuuCi about 1 month ago - 1 comment

#1227 - Make config/class properties on ComposerMPTForCausalLM

Pull Request - State: closed - Opened by dakinggg about 1 month ago

#1226 - Loss v len callback

Pull Request - State: closed - Opened by ShashankMosaicML about 1 month ago

#1225 - Add user error superclass

Pull Request - State: closed - Opened by milocress about 1 month ago - 2 comments

#1224 - Modularize components of megablocks layer builder

Pull Request - State: closed - Opened by dakinggg about 1 month ago - 3 comments

#1223 - Decompression tokens

Pull Request - State: closed - Opened by milocress about 1 month ago

#1222 - add error when chat template fails

Pull Request - State: closed - Opened by milocress about 1 month ago

#1221 - Finetuning does not work on nightly

Issue - State: closed - Opened by eldarkurtic about 1 month ago - 2 comments
Labels: bug

#1220 - Conversion Sharded -> Monolithic checkpoint

Issue - State: open - Opened by pretidav about 1 month ago - 1 comment
Labels: question

#1219 - Update readme to clarify flash-attn and TE installs

Pull Request - State: closed - Opened by snarayan21 about 1 month ago

#1218 - Add example eval scripts for dbrx PT sizes

Pull Request - State: closed - Opened by aspfohl about 1 month ago - 2 comments

#1217 - Add te for torch 2.4.0

Pull Request - State: closed - Opened by j316chuck about 1 month ago

#1216 - Fix dmoe tests GPU OOM

Pull Request - State: closed - Opened by snarayan21 about 1 month ago

#1215 - Update Dockerfile

Pull Request - State: closed - Opened by j316chuck about 1 month ago

#1214 - Dbfs HF

Pull Request - State: open - Opened by KuuCi about 1 month ago - 3 comments

#1213 - Removed debugging code in tests

Pull Request - State: closed - Opened by dakinggg about 1 month ago

#1212 - Adding HF source Path in for DBFS

Pull Request - State: closed - Opened by KuuCi about 1 month ago

#1210 - Added torch_dmoe defaults, bug fixes for 2D inputs

Pull Request - State: closed - Opened by snarayan21 about 1 month ago - 1 comment

#1209 - Add fc to HF export

Pull Request - State: closed - Opened by dakinggg about 1 month ago

#1208 - Update setup.py

Pull Request - State: closed - Opened by j316chuck about 1 month ago

#1207 - Chuck/gpu build te win

Pull Request - State: closed - Opened by j316chuck about 1 month ago - 1 comment

#1206 - Build Te

Pull Request - State: closed - Opened by j316chuck about 1 month ago

#1205 - Mvpatel2000/te image stable

Pull Request - State: closed - Opened by mvpatel2000 about 1 month ago

#1204 - TransformerEngine Image Build

Pull Request - State: closed - Opened by mvpatel2000 about 1 month ago

#1203 - [don't merge lol] test pyhook

Pull Request - State: closed - Opened by milocress about 1 month ago

#1202 - Clearer error message for unknown example type

Pull Request - State: closed - Opened by milocress about 2 months ago

#1201 - Make `fc_type` a dict to pass fc kwargs through

Pull Request - State: closed - Opened by snarayan21 about 2 months ago - 1 comment

#1200 - commit change

Pull Request - State: closed - Opened by j316chuck about 2 months ago

#1199 - Allow EOS token for finetuning

Pull Request - State: closed - Opened by jimwu6 about 2 months ago - 3 comments

#1198 - Removing rich install

Pull Request - State: closed - Opened by jjanezhang about 2 months ago

#1197 - MoE with FSDP

Issue - State: closed - Opened by Muennighoff about 2 months ago - 1 comment
Labels: question