Hugging Face's logo
wing lian's picture
68 8 449

wing lian PRO

winglian
anadi45's profile picture Qubitium's profile picture pwg2's profile picture
·

AI & ML interests

None yet

Organizations

Open Access AI Collective's profile picture Open-Source AI Meetup's profile picture Alignment Lab Ai's profile picture OpenChat's profile picture OpenOrca's profile picture AibTech's profile picture Open Phi's profile picture Qwen's profile picture 2 Year Future Ventures's profile picture textbooks's profile picture ZeroGPU Explorers's profile picture Wynd Network's profile picture Axolotl AI's profile picture Amplified 's profile picture Social Post Explorers's profile picture Cognitive Computations's profile picture gg-tt's profile picture Meta Llama's profile picture Axolotl AI Internal Models's profile picture Hugging Face Party @ PyTorch Conference's profile picture Axolotl AI's profile picture rg-preview's profile picture

winglian's activity

New activity in deepseek-ai/DeepSeek-Prover-V1.5-Base 23 days ago

the config class and config.json uses DeepseekConfig, not v2

1
#5 opened 23 days ago by winglian

Match the config class name to what the modeling code expects

1
#4 opened 23 days ago by winglian
New activity in microsoft/Phi-3.5-mini-instruct 28 days ago

trust_remote_code=True

1
#9 opened 28 days ago by winglian
New activity in NousResearch/Hermes-2-Pro-Llama-3-8B 5 months ago

add axolotl tag
#1 opened 5 months ago by winglian
New activity in mattshumer/Llama-3-8B-16K 5 months ago

add axolotl tag
#3 opened 5 months ago by winglian
New activity in cognitivecomputations/dolphin-2.9-llama3-8b 5 months ago

add axolotl tag
#12 opened 5 months ago by winglian
New activity in openbmb/Eurus-RM-7b 5 months ago

Enable flash_attention_2 support since the underlying Mistral model supports it
#3 opened 5 months ago by winglian
New activity in meta-llama/Meta-Llama-3-8B 5 months ago

Rename original/tokenizer.model to tokenizer.model

3
#6 opened 5 months ago by winglian
commented a paper 6 months ago
New activity in PrunaAI/dbrx-base-bnb-4bit 6 months ago

invalid weights doesn't match modeling code

1
#3 opened 6 months ago by winglian
New activity in SinclairSchneider/dbrx-base-quantization-fixed 6 months ago

reduce verbosity of logging
#1 opened 6 months ago by winglian
New activity in databricks/dbrx-instruct 6 months ago

The fused expert parameters means load_in_4bit doesn't work properly, nor does LoRA

31
#10 opened 6 months ago by tdrussell
New activity in LnL-AI/dbrx-base-converted-v2 6 months ago

reduce logging verbosity

1
#3 opened 6 months ago by winglian
New activity in SinclairSchneider/dbrx-instruct-quantization-fixed 6 months ago

dbrx-base

1
#2 opened 6 months ago by winglian
New activity in ai21labs/Jamba-v0.1 6 months ago

finetuning issues

2
#9 opened 6 months ago by winglian

Fix bias logic to enable QLoRA finetuning

3
#5 opened 6 months ago by winglian
New activity in cerebras/SlimPajama-627B 9 months ago

Trouble with streaming

6
#5 opened about 1 year ago by andersonbcdefg
New activity in open-llm-leaderboard/open_llm_leaderboard 10 months ago

latest commit breaks ability to submit mistral finetunes

4
#410 opened 10 months ago by winglian
New activity in Open-Orca/Mistral-7B-OpenOrca 10 months ago

Can you share the training configuration of Axolotl?

3
#24 opened 10 months ago by timlim123
New activity in open-llm-leaderboard/open_llm_leaderboard 10 months ago

possible model failure?

1
#403 opened 10 months ago by winglian

please remove openaccess-ai-collective/grendel

3
#387 opened 10 months ago by winglian

Unable to submit public models for evaluation

1
#379 opened 10 months ago by winglian
New activity in Open-Orca/SlimOrca-Dedup 11 months ago

minhash deduping

1
#2 opened 11 months ago by winglian
New activity in crumb/c4-benchfilter-nano 11 months ago

Script to run against more of c4?
#2 opened 11 months ago by winglian
New activity in stabilityai/stablelm-3b-4e1t 12 months ago

fix get_input_embdeddings

1
#3 opened 12 months ago by winglian
New activity in microsoft/phi-1_5 12 months ago

Attention mask not working during training

7
#34 opened 12 months ago by codegood
New activity in PygmalionAI/pygmalion-2-13b about 1 year ago

add axolotl badge to readme

1
#1 opened about 1 year ago by winglian
New activity in PygmalionAI/pygmalion-2-7b about 1 year ago

add axolotl badge to readme
#2 opened about 1 year ago by winglian
New activity in PygmalionAI/mythalion-13b about 1 year ago

Add axolotl badge to readme
#1 opened about 1 year ago by winglian
New activity in microsoft/phi-1_5 about 1 year ago

add _no_split_modules property
#17 opened about 1 year ago by winglian
New activity in garage-bAInd/Platypus2-13B about 1 year ago

Dataset

3
#1 opened about 1 year ago by winglian
New activity in Open-Orca/OpenOrcaxOpenChat-Preview2-13B about 1 year ago

Unreliable Benchmarks. Definitely worse than LLaMA2-13b

3
#5 opened about 1 year ago by anon7463435254
New activity in eugenepentland/oo-packing-checkpoint-15000 about 1 year ago

Upload 3 files
#1 opened about 1 year ago by winglian
New activity in winglian/t5-large-flan-cot about 1 year ago

Adding `safetensors` variant of this model
#1 opened about 1 year ago by SFconvertbot
New activity in openaccess-ai-collective/openllama-7b-4k over 1 year ago

What does the 4k stand for?

2
#1 opened over 1 year ago by flashvenom
New activity in openaccess-ai-collective/hippogriff-30b-chat over 1 year ago

Will you consider releasing a public dataset?

3
#1 opened over 1 year ago by Jackdiy

Set use_cache to True, otherwise inference performance is poor
#2 opened over 1 year ago by TheBloke
New activity in CarperAI/pythia-6.9b-deduped-4k over 1 year ago

inference with anything over 2k tokens causes the following error.
#1 opened over 1 year ago by winglian
New activity in mosaicml/mpt-7b over 1 year ago

upstream-replit-updates

4
#43 opened over 1 year ago by winglian
commented a paper over 1 year ago
New activity in TheBloke/WizardLM-30B-Uncensored-GPTQ over 1 year ago

Will this work with the Local LLMs One-Click UI runpod?

7
#2 opened over 1 year ago by nichedreams
New activity in openaccess-ai-collective/manticore-13b over 1 year ago

can you provie the data process demo before train llms?

2
#7 opened over 1 year ago by scall
New activity in openaccess-ai-collective/jeopardy-bot over 1 year ago

Token length 3908?

1
#1 opened over 1 year ago by Yhyu13
New activity in openaccess-ai-collective/StableLManticore-7B over 1 year ago

Is this based on LLaMA?

1
#1 opened over 1 year ago by Yhyu13
New activity in openaccess-ai-collective/lora-experiments-quant-to-full-weights over 1 year ago

What model is this?

1
#1 opened over 1 year ago by Yhyu13
New activity in openaccess-ai-collective/manticore-13b over 1 year ago

epoch 3 final? or 4 coming?

1
#6 opened over 1 year ago by faisalhr1997
New activity in BlinkDL/rwkv-4-pileplus over 1 year ago

14B
#1 opened over 1 year ago by winglian
New activity in openaccess-ai-collective/mpt-7b-wizardlm over 1 year ago

fine-tuning script notebook

4
#1 opened over 1 year ago by g30rv17ys
New activity in openaccess-ai-collective/manticore-13b over 1 year ago

Some suggestions for optimization

10
#3 opened over 1 year ago by polymer

specific instruct prompt to use

3
#2 opened over 1 year ago by digitous

What inputs does the model expect?

1
#4 opened over 1 year ago by AlmightYariv
New activity in openaccess-ai-collective/wizard-mega-13b over 1 year ago

Prompt format contradiction

2
#5 opened over 1 year ago by 2EyeGuy
New activity in openaccess-ai-collective/manticore-13b over 1 year ago

Difference with Wizzard Mega

1
#1 opened over 1 year ago by frandmb
New activity in openaccess-ai-collective/wizard-mega-13b over 1 year ago

Fine-tune specific details

6
#2 opened over 1 year ago by polymer
New activity in P1ayer-1/chatgpt-conversations-chatlogs.net over 1 year ago

v1 vs v2

1
#1 opened over 1 year ago by winglian
New activity in openaccess-ai-collective/ggml-ui over 1 year ago

Apply for community grant: Personal project

1
#1 opened over 1 year ago by winglian
New activity in theblackcat102/reward-deberta-v3-large-aspect over 1 year ago

training code
#1 opened over 1 year ago by winglian
New activity in openaccess-ai-collective/wizard-mega-13b over 1 year ago

Change cache = true in config.json to significantly boost inference performance

1
#1 opened over 1 year ago by TheBloke
New activity in mosaicml/mpt-7b over 1 year ago

Can this be fine-tuned with triton backed flash attention and alibi using the huggingface transformers trainer?

1
#13 opened over 1 year ago by winglian
New activity in TehVenom/MPT-7b-Chat-Instruct-LongCTX-Merge over 1 year ago

weighted average?

1
#4 opened over 1 year ago by winglian