Ecosyste.ms: Issues

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

GitHub / BlinkDL/ChatRWKV issues and pull requests

#107 - Convert vocabulary for the new tokenizer to JSON

Pull Request - State: open - Opened by koute almost 2 years ago

#106 - feat: add stopword checker + iterable generate function

Pull Request - State: open - Opened by Nintorac almost 2 years ago - 4 comments

#105 - Dev

Pull Request - State: closed - Opened by BlinkDL almost 2 years ago

#104 - Fix missing mm8_* methods in 'cuda fp16i8' when RWKV_CUDA_ON=0

Pull Request - State: closed - Opened by Blealtan almost 2 years ago

#103 - Struggling to get your custom CUDA kernel to run

Issue - State: closed - Opened by jojac47 almost 2 years ago - 4 comments

#102 - 关于训练数据集

Issue - State: closed - Opened by Galaxy-Ding almost 2 years ago - 9 comments

#101 - 你的胆量见识决定该项目根本没有什么前途

Issue - State: closed - Opened by akeliduo almost 2 years ago - 13 comments

#100 - Fix fp32i8 when RWKV_CUDA_ON.

Pull Request - State: closed - Opened by Blealtan almost 2 years ago - 4 comments

#98 - GPTQ for RWKV

Pull Request - State: open - Opened by 3outeille almost 2 years ago - 10 comments

#97 - RuntimeError: Error building extension 'wkv_cuda'

Issue - State: closed - Opened by linkerlin almost 2 years ago - 2 comments

#96 - 'RecursiveScriptModule' object has no attribute 'cuda_att_seq_i8'

Issue - State: closed - Opened by webpolis almost 2 years ago - 1 comment

#95 - feat: chat completion api inspired by openai

Pull Request - State: open - Opened by evshiron almost 2 years ago

#93 - [Solved]MBP x86 cpu(not M1/M2) + eGPU[rx6800 16G], can load model, but cannt work.

Issue - State: closed - Opened by FreeBlues almost 2 years ago - 3 comments

#92 - Add Japanese prompt

Pull Request - State: closed - Opened by tosiyuki almost 2 years ago - 1 comment

#91 - Can you combine ChatRWKV with ChatGLM-6B and Open-Assistant ?

Issue - State: open - Opened by anon62 almost 2 years ago - 1 comment

#90 - ModuleNotFoundError: No module named 'tokenizers.tokenizers'

Issue - State: closed - Opened by LouisYYC almost 2 years ago - 5 comments

#89 - update of rwkv error

Issue - State: closed - Opened by bello7777 almost 2 years ago - 2 comments

#88 - [a disappointing badcase on online demo]

Issue - State: closed - Opened by SrWYG almost 2 years ago - 3 comments

#87 - Chat state recovery

Pull Request - State: open - Opened by cryscan almost 2 years ago

#86 - How to change output behaivor?

Issue - State: open - Opened by sherry0429 almost 2 years ago - 2 comments

#85 - AssertionError: Torch not compiled with CUDA enabled

Issue - State: closed - Opened by wlp233 almost 2 years ago - 1 comment

#84 - what context length was the model trained

Issue - State: closed - Opened by snapo almost 2 years ago - 2 comments

#83 - Update model.py

Pull Request - State: closed - Opened by rushichavda almost 2 years ago - 1 comment

#82 - Update pip package RWKV.model and v2/chat.py to support LoRA

Pull Request - State: open - Opened by RafaRed almost 2 years ago - 4 comments

#81 - load new prompt while running

Pull Request - State: closed - Opened by zk-wz almost 2 years ago

#80 - Make work with multiple gpus

Pull Request - State: closed - Opened by BlinkDL almost 2 years ago

#79 - Argparse args

Pull Request - State: closed - Opened by hypnopump almost 2 years ago

#78 - RuntimeError: CUDA error: an illegal memory access was encountered

Issue - State: closed - Opened by nenkoru almost 2 years ago - 4 comments

#77 - feat: add windows paltform demo

Pull Request - State: closed - Opened by JoySeeDog almost 2 years ago

#76 - Tutorial chatv2

Pull Request - State: closed - Opened by elsatch almost 2 years ago

#75 - Updated requirements.txt to match V2 requirements

Pull Request - State: open - Opened by elsatch almost 2 years ago

#74 - Test quality with int8 and int4

Issue - State: closed - Opened by Jawad1347 almost 2 years ago - 1 comment

#73 - Simplify v2/Chat.py

Issue - State: closed - Opened by djaffer almost 2 years ago - 8 comments

#72 - Adds instructions for CPU-only 'cpu bf16' mode

Pull Request - State: closed - Opened by balisujohn almost 2 years ago

#71 - Maximum Token Length

Issue - State: closed - Opened by bitnom almost 2 years ago - 1 comment

#70 - The model BlinkDL/rwkv-4-pile-14b runs out of memory on HF with 32 GB RAM CPU

Issue - State: closed - Opened by djaffer almost 2 years ago - 12 comments

#69 - Improve code/instruction for how to download model fsx/BlinkDL/HF-MODEL

Issue - State: closed - Opened by djaffer almost 2 years ago - 1 comment

#68 - Fix vocab padding.

Pull Request - State: closed - Opened by cryscan almost 2 years ago

#67 - Severe output quality difference between 4096 and 8192 models

Issue - State: closed - Opened by GenTxt almost 2 years ago - 5 comments

#66 - Hi any example of conversation ?

Issue - State: closed - Opened by djaffer almost 2 years ago - 1 comment

#65 - how to make it run on Apple silicon(M1) on Mac

Issue - State: closed - Opened by eisneim almost 2 years ago - 5 comments

#62 - failed to compile wkv_cuda

Issue - State: closed - Opened by jcl2023 almost 2 years ago - 3 comments

#61 - Format CUDA optimized MM8 Kernel

Pull Request - State: closed - Opened by cryscan almost 2 years ago

#60 - Compile issue on Linux

Issue - State: closed - Opened by jasonmhead almost 2 years ago - 1 comment

#59 - Expanding more on running on Windows with CUDA

Issue - State: closed - Opened by jasonmhead almost 2 years ago - 1 comment

#58 - feat: docker support

Pull Request - State: open - Opened by soulteary almost 2 years ago - 3 comments

#57 - Add prompt_toolkit to requirements.txt

Issue - State: closed - Opened by leszekhanusz almost 2 years ago - 1 comment

#56 - v2 models

Issue - State: closed - Opened by loretoparisi almost 2 years ago - 1 comment

#55 - Update prompt.

Pull Request - State: closed - Opened by cryscan almost 2 years ago

#53 - Add optimized cuda mm8_one and pad vocab size to multiple of 64

Pull Request - State: closed - Opened by johanwind almost 2 years ago

#52 - Dev

Pull Request - State: closed - Opened by BlinkDL almost 2 years ago

#51 - Replace dots on loading with numbers

Pull Request - State: open - Opened by EgrorBs almost 2 years ago

#50 - Dev

Pull Request - State: closed - Opened by BlinkDL almost 2 years ago

#49 - sync

Pull Request - State: closed - Opened by BlinkDL almost 2 years ago

#48 - Validate strategy with regex

Pull Request - State: closed - Opened by EgrorBs almost 2 years ago - 1 comment

#47 - Allow loading and saving preconverted models.

Pull Request - State: closed - Opened by KerfuffleV2 almost 2 years ago - 11 comments

#46 - Remove pytorch dependency

Issue - State: closed - Opened by oobabooga almost 2 years ago - 2 comments

#45 - prompt

Pull Request - State: closed - Opened by BlinkDL almost 2 years ago

#44 - Dev

Pull Request - State: closed - Opened by BlinkDL almost 2 years ago

#43 - Split large input context token array in slices

Pull Request - State: closed - Opened by EgrorBs almost 2 years ago

#42 - Use fp32 atomicAdd in one mode kernel.

Pull Request - State: closed - Opened by Blealtan almost 2 years ago - 3 comments

#41 - DefaultCPUAllocator: not enough memory

Issue - State: closed - Opened by 936384885xy almost 2 years ago - 4 comments

#40 - Speeding up loading/decreasing memory use with mmap

Issue - State: closed - Opened by KerfuffleV2 almost 2 years ago - 11 comments

#39 - Does the all the model weights use Apache-2.0 license?

Issue - State: closed - Opened by Green-li almost 2 years ago - 1 comment

#38 - 0.5.0 operators.cu fails to compile on compute 6.x

Issue - State: closed - Opened by KerfuffleV2 almost 2 years ago - 15 comments

#37 - prompt update

Pull Request - State: closed - Opened by BlinkDL almost 2 years ago

#36 - Dev

Pull Request - State: closed - Opened by BlinkDL almost 2 years ago

#35 - [Feature request] 4bit/3bit quantization

Issue - State: open - Opened by kabachuha almost 2 years ago - 3 comments

#34 - Implement a CUDA kernel for fp16i8.

Pull Request - State: closed - Opened by Blealtan almost 2 years ago

#33 - 部分变量放JSON

Pull Request - State: closed - Opened by HisAtri almost 2 years ago

#32 - add c++ extension for channel_mixing

Pull Request - State: closed - Opened by mingfeima almost 2 years ago - 2 comments

#31 - CUDA load error on Windows 10

Issue - State: closed - Opened by morgankohler almost 2 years ago - 2 comments

#30 - cpu fp32i8 for low RAM usage on CPU?

Issue - State: closed - Opened by Crataco almost 2 years ago - 10 comments

#29 - setup of ChatRWKV

Issue - State: closed - Opened by bello7777 almost 2 years ago - 6 comments

#28 - Add top_k parameter

Pull Request - State: closed - Opened by oobabooga almost 2 years ago - 1 comment

#27 - No time_shift use in ChatRWKV?

Issue - State: closed - Opened by 3outeille almost 2 years ago - 1 comment

#26 - Add dependencies

Pull Request - State: closed - Opened by troilus-canva almost 2 years ago

#25 - Making this library more like Hugging Face

Issue - State: closed - Opened by oobabooga almost 2 years ago - 3 comments

#24 - Repeated sentences (and a thank you)

Issue - State: closed - Opened by pwsoft almost 2 years ago - 25 comments

#23 - include "timeit" module in benchmark.py

Issue - State: open - Opened by bennmann almost 2 years ago - 2 comments

#22 - lowering temperature in chat.py leads to error.

Issue - State: closed - Opened by dbddv01 almost 2 years ago - 1 comment

#21 - Still OOM for v2

Issue - State: closed - Opened by kizunasunhy almost 2 years ago - 2 comments

#20 - 训练

Issue - State: closed - Opened by ZTurboX almost 2 years ago - 1 comment

#18 - I found a minor bug in v2 model.py strategy code

Pull Request - State: closed - Opened by tiendung almost 2 years ago - 1 comment

#15 - AMD Ubuntu 22.04 GPU - guide

Issue - State: open - Opened by bennmann about 2 years ago - 7 comments

#13 - How to train a ChatRWKV model on domain-specific data

Issue - State: closed - Opened by Just4Uzi about 2 years ago - 2 comments

#12 - Run Single Model on Multiple GPUs

Pull Request - State: closed - Opened by mpskex about 2 years ago - 6 comments

#5 - M1 Max MPS F32 / F16 Issues

Issue - State: closed - Opened by okpatil4u about 2 years ago - 3 comments