Ecosyste.ms: Issues

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

GitHub / Lightning-AI/litgpt issues and pull requests

#99 - `max_returned_tokens` as a generate input

Pull Request - State: closed - Opened by carmocca over 1 year ago - 2 comments

#98 - Update howto for finetuning on MPS

Pull Request - State: closed - Opened by carmocca over 1 year ago

#97 - doc: add `pip install huggingface_hub` to howtos

Pull Request - State: closed - Opened by emilesilvis over 1 year ago

#96 - Should `huggingface_hub` be added to requirements.txt?

Issue - State: closed - Opened by emilesilvis over 1 year ago - 5 comments

#95 - Query Regarding Minimum Hardware Requirements for Fine-tuning and Inference

Issue - State: closed - Opened by martinb-ai over 1 year ago - 6 comments

#93 - Fix an assertion during training issue #65

Pull Request - State: closed - Opened by ArturK-85 over 1 year ago - 5 comments

#92 - tpu howto timing adjustments

Pull Request - State: closed - Opened by gkroiz over 1 year ago

#91 - Reset all caches on xla devices

Pull Request - State: closed - Opened by gkroiz over 1 year ago - 1 comment

#90 - Adding padding preprocessing

Pull Request - State: closed - Opened by gkroiz over 1 year ago - 3 comments

#89 - Reorder adapter and subclass CausalSelfAttention

Pull Request - State: closed - Opened by carmocca over 1 year ago

#88 - Revert "Adding `vocab_size` to `config.py` (#68)"

Pull Request - State: closed - Opened by carmocca over 1 year ago - 6 comments

#86 - Restructure repo into directories as Lit-LLaMA

Pull Request - State: closed - Opened by carmocca over 1 year ago - 3 comments

#85 - Port Adapter_V2 from lit-llama pull request #303

Pull Request - State: closed - Opened by ArturK-85 over 1 year ago - 5 comments

#84 - Add chat script for adapter checkpoints

Issue - State: closed - Opened by RDouglasSharp over 1 year ago - 3 comments
Labels: enhancement, generation

#83 - reset cache after generation

Pull Request - State: closed - Opened by bkiat1123 over 1 year ago - 7 comments

#82 - Add adapter multihead gating

Pull Request - State: closed - Opened by rasbt over 1 year ago - 1 comment

#81 - Updated TPU docs to download nightly torch dependencies

Pull Request - State: closed - Opened by gkroiz over 1 year ago

#80 - Do not use symlinks when downloading for cloud support

Pull Request - State: closed - Opened by carmocca over 1 year ago

#79 - Fix max_iters logic

Pull Request - State: closed - Opened by carmocca over 1 year ago

#78 - Model finetuned using finetune_adapter not directly usable in generaete/chat... How to convert?

Issue - State: closed - Opened by RDouglasSharp over 1 year ago - 3 comments
Labels: enhancement, generation

#77 - Caches should not persist across multiple generate.

Issue - State: closed - Opened by bkiat1123 over 1 year ago - 1 comment
Labels: enhancement, generation

#76 - fix typo

Pull Request - State: closed - Opened by bkiat1123 over 1 year ago

#75 - fix unpack issue

Pull Request - State: closed - Opened by bkiat1123 over 1 year ago

#74 - fix unpacking error

Pull Request - State: closed - Opened by bkiat1123 over 1 year ago - 1 comment

#73 - too many values to unpack in Block forward

Issue - State: closed - Opened by bkiat1123 over 1 year ago

#72 - fix quantization error

Pull Request - State: closed - Opened by bkiat1123 over 1 year ago - 2 comments

#71 - Add T_new as an argument in generate, reference in issue #65 #69

Pull Request - State: closed - Opened by ArturK-85 over 1 year ago

#70 - Add T_max as an argument in generate, reference to #65

Pull Request - State: closed - Opened by ArturK-85 over 1 year ago

#69 - Add adapter tests

Issue - State: closed - Opened by carmocca over 1 year ago

#68 - Adding `vocab_size` to `config.py`

Pull Request - State: closed - Opened by gkroiz over 1 year ago

#67 - Update generate.py

Pull Request - State: closed - Opened by ArturK-85 over 1 year ago

#66 - Adding caching for adapter KVs

Pull Request - State: closed - Opened by gkroiz over 1 year ago

#65 - Assert in generate.py needs to go...

Issue - State: closed - Opened by RDouglasSharp over 1 year ago - 5 comments

#64 - Set dtype of kv_caches correctly

Pull Request - State: closed - Opened by lantiga over 1 year ago

#63 - Port packed_dataset hotfix from lit-llama

Pull Request - State: closed - Opened by lantiga over 1 year ago

#62 - Fine tune adapter device = 2 deepspeed error

Issue - State: closed - Opened by ArturK-85 over 1 year ago - 1 comment
Labels: bug

#61 - Problem with finetune_adapter.py along with fix

Issue - State: closed - Opened by RDouglasSharp over 1 year ago - 8 comments

#60 - Apply proposed changes by @RDouglasSharp from issue #57 #58

Pull Request - State: closed - Opened by ArturK-85 over 1 year ago - 1 comment

#59 - Config cannot be overwritten through kwargs

Issue - State: closed - Opened by bkiat1123 over 1 year ago - 2 comments

#58 - Cached KVs not implemented on Adapter causing errors.

Issue - State: closed - Opened by bkiat1123 over 1 year ago - 2 comments

#57 - /lit_parrot/model.py:201 in forward

Issue - State: closed - Opened by RDouglasSharp over 1 year ago - 2 comments

#56 - Cleanup reply output in chat.py

Pull Request - State: closed - Opened by gkroiz over 1 year ago

#55 - Allow generate seed

Pull Request - State: closed - Opened by Sciumo over 1 year ago

#54 - Generate should allow seed, rather than setting to 1234.

Issue - State: closed - Opened by Sciumo over 1 year ago - 5 comments

#53 - Get Attempting to unscale FP16 gradients while finetune on float16

Issue - State: closed - Opened by bkiat1123 over 1 year ago - 7 comments

#52 - Fix n_embd for larger Pythia models

Pull Request - State: closed - Opened by carmocca over 1 year ago

#51 - Adding cached KVs

Pull Request - State: closed - Opened by gkroiz over 1 year ago - 2 comments

#50 - Pythia embedding dimension mismatch

Issue - State: closed - Opened by rcmalli over 1 year ago - 1 comment

#49 - Fix training argument

Pull Request - State: closed - Opened by lantiga over 1 year ago

#48 - fix unrecognized command error

Pull Request - State: closed - Opened by aniketmaurya over 1 year ago - 3 comments

#47 - Add generation and chat scripts support for XLA devices

Pull Request - State: closed - Opened by gkroiz over 1 year ago

#46 - change checkpoint dir test to renable convert to Parrot (but still check)

Pull Request - State: closed - Opened by Sciumo over 1 year ago - 3 comments

#45 - Restore generation longer than block size

Pull Request - State: closed - Opened by carmocca over 1 year ago

#44 - Download documentation needs updating, --repo_id required

Issue - State: closed - Opened by Sciumo over 1 year ago - 3 comments

#42 - Fix model configuration

Pull Request - State: closed - Opened by rcmalli over 1 year ago

#41 - Remove checkpoint_dir check when converting weights

Pull Request - State: closed - Opened by carmocca over 1 year ago

#40 - Minor train_redpajama fixes

Pull Request - State: closed - Opened by carmocca over 1 year ago

#39 - Fix max_new_tokens logic

Pull Request - State: closed - Opened by carmocca over 1 year ago

#38 - Parrot rename

Pull Request - State: closed - Opened by lantiga over 1 year ago

#37 - ERROR: Could not find a version that satisfies the requirement torch>=2.1.0dev

Issue - State: closed - Opened by nestorvw over 1 year ago - 2 comments

#36 - Updated README and cleaned up howto

Pull Request - State: closed - Opened by lantiga over 1 year ago - 1 comment

#35 - Port pre-training scripts

Pull Request - State: closed - Opened by lantiga over 1 year ago

#34 - Update finetune_adapter.py

Pull Request - State: closed - Opened by awaelchli over 1 year ago

#33 - Update howto for finetuning with adapter

Pull Request - State: closed - Opened by awaelchli over 1 year ago

#32 - Update download instructions

Pull Request - State: closed - Opened by awaelchli over 1 year ago

#31 - Porting adapter from lit-llama

Pull Request - State: closed - Opened by ArturK-85 over 1 year ago - 5 comments

#30 - lit-stablelm/finetune_adapter.py

Pull Request - State: closed - Opened by ArturK-85 over 1 year ago

#29 - lit-stablelm/adapter

Pull Request - State: closed - Opened by ArturK-85 over 1 year ago

#28 - Better UX for discovering checkpoints

Pull Request - State: closed - Opened by carmocca over 1 year ago

#27 - Add mascots to README

Pull Request - State: closed - Opened by lantiga over 1 year ago

#26 - Split configs into a separate file

Pull Request - State: closed - Opened by carmocca over 1 year ago

#25 - Add INCITE howto

Pull Request - State: closed - Opened by carmocca over 1 year ago

#24 - Support INCITE checkpoints

Pull Request - State: closed - Opened by carmocca over 1 year ago

#23 - Single checkpoint dir to simplify script arguments

Pull Request - State: closed - Opened by carmocca over 1 year ago - 1 comment

#22 - Add interactive chatting script

Pull Request - State: closed - Opened by carmocca over 1 year ago

#21 - KV cache for faster generation

Issue - State: closed - Opened by carmocca over 1 year ago
Labels: enhancement, generation

#20 - Port finetuning from Lit-LLaMA

Issue - State: closed - Opened by carmocca over 1 year ago - 2 comments
Labels: enhancement, fine-tuning

#19 - Add Pythia howto

Pull Request - State: closed - Opened by carmocca over 1 year ago

#18 - Quantization is not working

Issue - State: closed - Opened by carmocca over 1 year ago - 2 comments
Labels: bug, quantization

#17 - Add gif animation

Pull Request - State: closed - Opened by lantiga over 1 year ago

#16 - Update memory usage in docs

Pull Request - State: closed - Opened by carmocca over 1 year ago

#15 - Add all StableLM and Pythia configs

Pull Request - State: closed - Opened by carmocca over 1 year ago

#14 - Which should be our default model?

Issue - State: closed - Opened by carmocca over 1 year ago - 2 comments
Labels: documentation

#13 - RoPE precision issue

Issue - State: closed - Opened by carmocca over 1 year ago - 3 comments
Labels: bug

#12 - Load smaller model in generation test

Pull Request - State: closed - Opened by carmocca over 1 year ago

#11 - Install torch nightly for flash attention

Pull Request - State: closed - Opened by carmocca over 1 year ago

#10 - Fix CPU OOM on Windows

Issue - State: closed - Opened by carmocca over 1 year ago

#9 - Improve UX for discovering available checkpoints

Issue - State: closed - Opened by carmocca over 1 year ago - 1 comment
Labels: enhancement

#8 - Support all StableLM and Pythia checkpoint configs

Issue - State: closed - Opened by carmocca over 1 year ago

#7 - Write Pythia checkpoint howto

Issue - State: closed - Opened by carmocca over 1 year ago
Labels: documentation

#6 - Support tuned-model mode during generation

Issue - State: closed - Opened by carmocca over 1 year ago - 1 comment
Labels: enhancement

#5 - Support Pythia checkpoints

Pull Request - State: closed - Opened by carmocca over 1 year ago - 3 comments

#4 - Add README

Pull Request - State: closed - Opened by lantiga over 1 year ago

#3 - Fix generate.py on Windows

Pull Request - State: closed - Opened by carmocca over 1 year ago

#2 - Flash attention support

Issue - State: closed - Opened by carmocca over 1 year ago - 2 comments

#1 - Initial commit

Pull Request - State: closed - Opened by carmocca over 1 year ago - 1 comment