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
#299 - Default to debug level debug
Pull Request -
State: closed - Opened by samhavens over 1 year ago
#294 - Update README.md
Pull Request -
State: closed - Opened by abhi-mosaic over 1 year ago
- 2 comments
#292 - Upgrade to `mosaicml-streaming==0.5.x`
Pull Request -
State: closed - Opened by abhi-mosaic over 1 year ago
- 3 comments
#290 - Constant training loss observed when using mpt-7b_dolly_sft.yaml config
Issue -
State: closed - Opened by suehyunpark over 1 year ago
- 5 comments
#288 - Add shift_labels arg to HF wrappers
Pull Request -
State: closed - Opened by dakinggg over 1 year ago
- 1 comment
#285 - Small formatting fix in eval README
Pull Request -
State: closed - Opened by sashaDoubov over 1 year ago
#281 - Can't produce same answer
Issue -
State: closed - Opened by apachemycat over 1 year ago
- 1 comment
Labels: question
#279 - Add 8-bit LION optimizer
Pull Request -
State: closed - Opened by dblalock over 1 year ago
- 4 comments
#278 - Adding custom embedding
Pull Request -
State: closed - Opened by bcui19 over 1 year ago
- 1 comment
#271 - adding te Linear for fp8 support
Pull Request -
State: closed - Opened by vchiley over 1 year ago
- 8 comments
#264 - Model loading on local machine
Issue -
State: closed - Opened by Devangkaruskar over 1 year ago
- 6 comments
Labels: question
#260 - updt tritonpremlir to sm90 version
Pull Request -
State: closed - Opened by vchiley over 1 year ago
- 4 comments
#259 - Multiple models inference on Single-GPU
Issue -
State: closed - Opened by gitsand996 over 1 year ago
- 1 comment
Labels: question
#258 - ERROR:composer.cli.launcher:Rank 2 crashed with exit code -7
Issue -
State: closed - Opened by tb852 over 1 year ago
- 3 comments
#248 - Configure eval to give 'loss/eval' that is analgous to 'loss/train'
Issue -
State: closed - Opened by tginart over 1 year ago
- 6 comments
#247 - ERROR: expected to be in states [<TrainingState_.IDLE: 1>] but current state is TrainingState_.BACKWARD_PRE
Issue -
State: closed - Opened by NarenZen over 1 year ago
- 4 comments
#245 - Generated sample equals to input samples
Issue -
State: closed - Opened by germanjke over 1 year ago
- 4 comments
#243 - Multi-nodes slurm training
Issue -
State: closed - Opened by j-Gaow over 1 year ago
- 2 comments
#240 - Fix Typing (part 1)
Pull Request -
State: closed - Opened by hanlint over 1 year ago
#234 - Refactor logging
Pull Request -
State: closed - Opened by hanlint over 1 year ago
- 2 comments
#224 - timeout error
Issue -
State: closed - Opened by NarenZen over 1 year ago
- 4 comments
#223 - Why does convert_dataset_json.py only support 'train' for the --split argument?"
Issue -
State: closed - Opened by sysusicily over 1 year ago
- 1 comment
#217 - "triton-pre-mlir" installation issue using the command pip install -e ".[gpu]"
Issue -
State: closed - Opened by satyaskada over 1 year ago
- 7 comments
#214 - G-Eval
Pull Request -
State: closed - Opened by samhavens over 1 year ago
- 1 comment
#212 - Torch 1.13.1 doesn't support sm_90
Issue -
State: open - Opened by jwatte over 1 year ago
#211 - Question
Issue -
State: open - Opened by ChiefBlacktail over 1 year ago
#210 - Kv cache speed
Pull Request -
State: closed - Opened by vchiley over 1 year ago
#209 - Use $RUN_NAME rather than $COMPOSER_RUN_NAME
Pull Request -
State: open - Opened by abhi-mosaic over 1 year ago
#208 - Refresh Mosaicml platform yamls
Pull Request -
State: open - Opened by aspfohl over 1 year ago
#207 - Update README.md - Slack Link
Pull Request -
State: open - Opened by ejyuen over 1 year ago
#206 - Removed unused `tokenizer_name` config field
Pull Request -
State: closed - Opened by dakinggg over 1 year ago
#205 - Onboarding tutorial and related improvements
Pull Request -
State: open - Opened by alextrott16 over 1 year ago
#204 - Update inference README
Pull Request -
State: closed - Opened by abhi-mosaic over 1 year ago
#203 - Error:"Watchdog caught collective operation timeout" when finetuning MPT-7B on a local dataset using 2 A100 GPUs
Issue -
State: closed - Opened by satyaskada over 1 year ago
- 15 comments
#202 - S3 ckpt saving
Issue -
State: open - Opened by germanjke over 1 year ago
- 1 comment
#201 - ONNX conversion is too memory expensive
Issue -
State: open - Opened by makermotion over 1 year ago
#200 - Providing an input context.
Issue -
State: closed - Opened by thusithaC over 1 year ago
- 2 comments
#199 - eos tokens
Issue -
State: open - Opened by tginart over 1 year ago
#198 - Update README.md
Pull Request -
State: closed - Opened by jacobfulano over 1 year ago
#197 - test ci
Pull Request -
State: closed - Opened by vchiley over 1 year ago
#196 - Kernel Crashes when trying to load model to CUDA
Issue -
State: open - Opened by souvik0306 over 1 year ago
- 2 comments
#195 - Inferencing with multigpu
Issue -
State: open - Opened by singhalshikha518 over 1 year ago
#194 - learning rate for pre-training
Issue -
State: closed - Opened by sysusicily over 1 year ago
- 1 comment
#193 - torch2 updt with hf fixes
Pull Request -
State: closed - Opened by vchiley over 1 year ago
- 1 comment
#192 - Tensor Parallel MLP with torch2.0
Pull Request -
State: closed - Opened by dskhudia over 1 year ago
- 7 comments
#191 - How to adapt to different context size?
Issue -
State: closed - Opened by jwatte over 1 year ago
- 1 comment
#190 - Triton not working on A40 and A6000 machines
Issue -
State: open - Opened by NarenZen over 1 year ago
- 2 comments
#189 - ERROR: Could not build wheels for flash-attn, xentropy-cuda-lib, which is required to install pyproject.toml-based projects on A6000 machine
Issue -
State: open - Opened by NarenZen over 1 year ago
- 2 comments
#188 - GPTQ support for quantization
Issue -
State: open - Opened by casperbh96 over 1 year ago
- 2 comments
#187 - downloading datasets
Issue -
State: open - Opened by germanjke over 1 year ago
- 1 comment
#186 - ALiBi with `causal=True` unexpected bias?
Issue -
State: open - Opened by KeremTurgutlu over 1 year ago
- 7 comments
#185 - unexpected results in inference
Issue -
State: open - Opened by OmarMohammed88 over 1 year ago
- 4 comments
#184 - WandB integration?
Issue -
State: closed - Opened by germanjke over 1 year ago
- 1 comment
#183 - GPU OOM while fine-tuning MPT-7B
Issue -
State: closed - Opened by karthikmurugadoss over 1 year ago
- 5 comments
#182 - Add community links to README
Pull Request -
State: closed - Opened by hanlint over 1 year ago
#181 - Revert "Torch2 (#177) (#178)"
Pull Request -
State: closed - Opened by dakinggg over 1 year ago
- 2 comments
#180 - Flash Attention vs Triton Flash Attention
Issue -
State: open - Opened by germanjke over 1 year ago
- 6 comments
#179 - MPT-7B strange inference speed
Issue -
State: closed - Opened by SinanAkkoyun over 1 year ago
- 1 comment
#178 - Torch2 (#177)
Pull Request -
State: closed - Opened by vchiley over 1 year ago
- 1 comment
#177 - Torch2
Pull Request -
State: closed - Opened by vchiley over 1 year ago
#176 - Error while loading converted hf model(from composer checkpoint)
Issue -
State: closed - Opened by singhalshikha518 over 1 year ago
- 4 comments
#175 - Rename datasets to avoid hf conflict
Pull Request -
State: closed - Opened by hanlint over 1 year ago
#174 - TypeError: init() got an unexpected keyword argument 'approximate' when using mosaicml/mpt-7b-instruct model
Issue -
State: closed - Opened by souvik0306 over 1 year ago
- 10 comments
#173 - Can I use mpt-7b_dolly_sft.yaml used to train MPT-Instruct model
Issue -
State: closed - Opened by NarenZen over 1 year ago
- 1 comment
#172 - Dynamic range of ALiBi
Issue -
State: closed - Opened by tginart over 1 year ago
- 1 comment
#171 - where can I find training code or configs about MPT-7B-StoryWriter
Issue -
State: open - Opened by metacryptom over 1 year ago
#170 - FileNotFoundError: [Errno 2] No such file or directory: '/000001_barrier'
Issue -
State: closed - Opened by julianfaraone over 1 year ago
- 5 comments
#169 - Convert MPT checkpoints to FT format
Pull Request -
State: closed - Opened by dskhudia over 1 year ago
- 2 comments
#168 - clean up dataset conversion readme
Pull Request -
State: closed - Opened by codestar12 over 1 year ago
#167 - Remove health checker
Pull Request -
State: closed - Opened by mvpatel2000 over 1 year ago
#166 - Add Tensorboard logger to yaml config
Pull Request -
State: closed - Opened by hanlint over 1 year ago
- 2 comments
#165 - Remove `pynvml`
Pull Request -
State: closed - Opened by hanlint over 1 year ago
- 1 comment
#164 - Explain `composer` command
Pull Request -
State: closed - Opened by hanlint over 1 year ago
#163 - Error while saving checkpoint
Issue -
State: closed - Opened by singhalshikha518 over 1 year ago
- 14 comments
#162 - ValueError: --out_root=./my-copy-c4 contains ['train_small'] which cannot overlap with the requested splits ['train_small', 'val_small'].
Issue -
State: closed - Opened by NishaDeepak over 1 year ago
- 3 comments
#161 - ValueError("Please specify `target_modules` in `peft_config`")
Issue -
State: closed - Opened by NarenZen over 1 year ago
- 1 comment
#160 - Finetune MPT with transformers
Issue -
State: closed - Opened by yangjianxin1 over 1 year ago
- 2 comments
#159 - an error while training
Issue -
State: open - Opened by ChrisXULC over 1 year ago
- 3 comments
#158 - support for tensorboard
Issue -
State: closed - Opened by sysusicily over 1 year ago
- 3 comments
#157 - Update StreamingDataset defaults
Pull Request -
State: closed - Opened by abhi-mosaic over 1 year ago
#156 - Adds a concrete finetuning example from a custom dataset
Pull Request -
State: closed - Opened by alextrott16 over 1 year ago
- 1 comment
#155 - Not getting a proper response
Issue -
State: closed - Opened by gauravkaliadev over 1 year ago
- 4 comments
#154 - Use the nvidia-supplied nvidia-ml-py instead of pynvml
Issue -
State: closed - Opened by mattip over 1 year ago
- 3 comments
#153 - Slow on V100
Issue -
State: open - Opened by Louis-y-nlp over 1 year ago
- 2 comments
#152 - Docker Image with CUDA 12.1 for ADA Gen cards
Issue -
State: open - Opened by danzeeeman over 1 year ago
- 2 comments
#151 - Add cloud upload to checkpoint conversion script
Pull Request -
State: closed - Opened by dakinggg over 1 year ago
#149 - Enable Torch2
Pull Request -
State: closed - Opened by vchiley over 1 year ago
- 12 comments
#148 - Adds precision to eval
Pull Request -
State: closed - Opened by mvpatel2000 over 1 year ago
#147 - make triton attn req pre-mlri tagged triton
Pull Request -
State: closed - Opened by vchiley over 1 year ago
- 2 comments
#146 - error 'Getting requirements to build wheel'... is the docker image okay??
Issue -
State: closed - Opened by jewbot over 1 year ago
- 7 comments
#144 - Explicit composer mention
Issue -
State: closed - Opened by StrangeTcy over 1 year ago
- 2 comments
#143 - fine tuning mpt7b using local dataset
Issue -
State: open - Opened by singhalshikha518 over 1 year ago
- 10 comments
#142 - the select for multi-GPU card
Issue -
State: closed - Opened by sysusicily over 1 year ago
- 1 comment
#141 - the error of streaming
Issue -
State: open - Opened by sysusicily over 1 year ago
- 3 comments
#140 - Why Flash Attention do not support attn bias [Alibi]?
Issue -
State: closed - Opened by srn-source over 1 year ago
- 2 comments
#139 - KeyError: 'attn_pdrop' with t5-small_dolly_sft.yaml when running inference/convert_composer_to_hf.py
Issue -
State: open - Opened by Paladiamors over 1 year ago
- 1 comment
#137 - Circular import error when using data/packing.py
Issue -
State: closed - Opened by Paladiamors over 1 year ago
- 2 comments
#132 - How to load a dataset with multiple rounds of conversation like sharegpt
Issue -
State: closed - Opened by 0xDing over 1 year ago
- 1 comment
#129 - code example for the onnx model
Issue -
State: closed - Opened by therealadityashankar over 1 year ago
- 4 comments
#128 - Unable to use triton? How to handle context windows >4k?
Issue -
State: closed - Opened by tginart over 1 year ago
- 2 comments