Ecosyste.ms: Issues

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

GitHub / RWKV/RWKV-infctx-trainer issues and pull requests

#101 - M8than fix variable length

Pull Request - State: open - Opened by m8than 8 months ago

#100 - Update model.py

Pull Request - State: closed - Opened by m8than 8 months ago

#99 - Rwkv x5 segmented loss

Pull Request - State: closed - Opened by PicoCreator 9 months ago

#98 - How do I build my own training datasets?

Issue - State: closed - Opened by cgoxopx 9 months ago

#96 - Variable length dataset fix, (0 length input fix)

Pull Request - State: closed - Opened by m8than 9 months ago

#95 - Fix trie-tokenizer decoding

Pull Request - State: closed - Opened by freckletonj 9 months ago - 4 comments

#94 - remove improper asserts

Pull Request - State: closed - Opened by SmerkyG 9 months ago

#93 - dtype support for v6

Pull Request - State: closed - Opened by SmerkyG 9 months ago

#92 - RWKV v6 tweaks

Pull Request - State: closed - Opened by PicoCreator 10 months ago

#91 - Porting back fixes to rwkv-x6

Pull Request - State: closed - Opened by PicoCreator 10 months ago

#90 - Rwkv x h100 main merge

Pull Request - State: closed - Opened by PicoCreator 10 months ago

#89 - pulling main updates to rwkv-x-playground

Pull Request - State: closed - Opened by PicoCreator 10 months ago

#88 - RWKV 6 support

Pull Request - State: closed - Opened by SmerkyG 10 months ago

#87 - Eugene/train test split

Pull Request - State: open - Opened by PicoCreator 10 months ago

#86 - Is the training of the RWKV-6 model not yet applicable?

Issue - State: open - Opened by wooks186 11 months ago - 1 comment

#85 - consolidated log, and datapack shuffling on trainer

Pull Request - State: closed - Opened by PicoCreator 11 months ago

#84 - Add the datapack shuffling, within the trainer

Pull Request - State: closed - Opened by PicoCreator 11 months ago

#81 - Rwkv x h100

Pull Request - State: closed - Opened by PicoCreator 11 months ago

#80 - jit improvement, noncuda fixes

Pull Request - State: closed - Opened by SmerkyG 11 months ago

#79 - Rwkv x maxjit

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#78 - Rwkv x eagle notebooks

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#77 - Rwkv x eagle notebooks

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#75 - Rwkv x eagle notebooks

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#74 - Update data.py

Pull Request - State: open - Opened by m8than 12 months ago

#73 - Rwkv x eagle notebooks

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#72 - Rwkv x eagle notebooks

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#71 - Rwkv x interweave datapack

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#70 - Rwkv nocuda

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#69 - run tweak

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#68 - datapacking support

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#66 - Remote dataset support

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#65 - Rwkv x selective loss exp

Pull Request - State: closed - Opened by PicoCreator 12 months ago

#64 - [Need help] Implement BPTT cuda code for V5

Issue - State: closed - Opened by PicoCreator about 1 year ago - 2 comments

#63 - Ds3 fix

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#62 - [Feature] Dataset streaming

Issue - State: closed - Opened by PicoCreator about 1 year ago - 1 comment

#61 - Rwkv x selective loss exp

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#60 - tweaking build rules

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#59 - Rwkv x selective loss exp

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#58 - sync-up playground

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#57 - Selective loss training

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#56 - Dataset packing

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#55 - Conversational data feature

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#54 - relu r, k

Pull Request - State: closed - Opened by SmerkyG about 1 year ago

#53 - relu r, k

Pull Request - State: closed - Opened by SmerkyG about 1 year ago

#52 - relu r, k

Pull Request - State: closed - Opened by SmerkyG about 1 year ago

#51 - relu r, k

Pull Request - State: closed - Opened by SmerkyG about 1 year ago

#50 - merge back into playground

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#49 - Fix config example

Pull Request - State: closed - Opened by m8than about 1 year ago

#48 - fix: loss forward segment count moved to correct device

Pull Request - State: closed - Opened by m8than about 1 year ago

#46 - V5 batching support

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#45 - V5 batching support + extras

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#44 - Full v5 r4 rewrite

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#43 - Update README.md

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#42 - Full v5 r4 rewrite

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#41 - Any pretrained models avaliable?

Issue - State: closed - Opened by BrightXiaoHan about 1 year ago

#40 - Rwkv x playground r4 h t

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#39 - fix for broken state management

Pull Request - State: closed - Opened by harrisonvanderbyl about 1 year ago

#38 - Syncing merged changes to rwkv-x-playground

Pull Request - State: closed - Opened by PicoCreator about 1 year ago

#37 - Fix `segment_count` not properly moved onto device

Pull Request - State: closed - Opened by TearGosling about 1 year ago

#36 - [Feature request]Please eliminate CUDA dependency

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

#35 - bug w.r.t. to multicolumn masking

Issue - State: closed - Opened by jprobichaud over 1 year ago - 4 comments

#34 - "\n" is escaped to "\\n" in preload_datapath.py

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

#33 - Add code for dataset generation

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

#32 - Add code to run math and name dataset

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

#31 - I edited the lightning_trainer after getting some errors

Pull Request - State: closed - Opened by Bruber92 over 1 year ago - 1 comment

#30 - [Feature] Run OpenAI compatible API server for eval harness

Issue - State: open - Opened by PicoCreator over 1 year ago - 1 comment

#29 - M8than patch 1

Pull Request - State: closed - Opened by m8than over 1 year ago

#28 - Optimizing model loading memory bump, for multi GPU training

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

#26 - [Feature] Add (kilo) tokens / second into the reporting

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

#25 - [Feature] Support for microbatches

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

#24 - No rechunking of text by default

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

#23 - Dataset Sorting + multi column suffix features

Pull Request - State: closed - Opened by m8than over 1 year ago

#22 - main branch sync

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

#21 - [Feature] positive / negative data support (aka RFHL)

Issue - State: open - Opened by PicoCreator over 1 year ago

#20 - [Feature] LoRA support

Issue - State: open - Opened by PicoCreator over 1 year ago - 1 comment

#19 - Main dev infctx

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

#18 - Added token 0 to the tokenizer.

Pull Request - State: closed - Opened by m8than over 1 year ago - 1 comment

#17 - limited dataloader num_worker max to 8

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

#16 - Unable to finetune official RWKV-5 models

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

#14 - Update README.md

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

#13 - Suggestion: option to only keep last.ckpt and auto delete older ones

Issue - State: closed - Opened by h-a-s-k over 1 year ago - 4 comments

#12 - fixing vocab size mismatch

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

#11 - fix accidental removal of ctx_len_cutoffs

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

#10 - Main dev infctx

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

#9 - revert

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

#8 - Main dev infctx

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

#7 - dataload做了全核心数的num_workers是否有必要

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

#6 - Main dev infctx

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

#5 - Main dev infctx

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

#4 - Main dev infctx

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

#3 - WIP - github action setup

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

#2 - roll up of multiple changes

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