Jawad Mansoor
supercharge19
AI & ML interests
NLP for text and voice (even videos)
RL with multimodaliy models (agent is able to learn human speech as well as can see and make decisions based on what it "sees")
Organizations
supercharge19's activity
Warning: llm_load_vocab: missing pre-tokenizer type, using: 'default'
1
#4 opened 1 day ago
by
supercharge19
The f16 with 32k ctx fits nicely in 24GB VRAM
5
#3 opened 5 days ago
by
ubergarm
Is there a plan to use llama3 ?
2
#11 opened 1 day ago
by
supercharge19
Oh Come on
6
#1 opened 2 days ago
by
supercharge19
What is 256k?
16
#1 opened 5 days ago
by
supercharge19
What is smashed?
2
#1 opened 4 days ago
by
supercharge19
Is it bitnet {-1,0,1}?
1
#3 opened 24 days ago
by
Remek
Is Hermes Pro 2 coming for this?
#2 opened 12 days ago
by
supercharge19
Why are these models fp32?
5
#2 opened about 1 month ago
by
supercharge19
Light vs original
#1 opened about 1 month ago
by
supercharge19
Add GGUF model file for llama.cpp (f16)
6
#3 opened about 2 months ago
by
iamlemec
is it fp16?
3
#1 opened about 1 month ago
by
supercharge19
Censored; generating gibberish... cannot get this model to function well.
13
#3 opened about 2 months ago
by
UniversalLove333
Is it censored?
5
#1 opened about 2 months ago
by
supercharge19
How is this created
6
#1 opened about 2 months ago
by
supercharge19
why not put deepseek coder in comparisons
#12 opened about 2 months ago
by
supercharge19
Good job
2
#2 opened about 2 months ago
by
supercharge19
Decending in ranking
2
#2 opened about 2 months ago
by
supercharge19
4K_0 gone and 4K_XS?
2
#2 opened about 2 months ago
by
supercharge19
Buggy GGUF Output
14
#38 opened 2 months ago
by
mattjcly
disclaimer 2
5
#1 opened 2 months ago
by
supercharge19
what are these experiments?
2
#1 opened 2 months ago
by
supercharge19
You're missing an opportunity to boost performance.
3
#3 opened 2 months ago
by
Phil337
Is instruction format necessary
2
#142 opened 2 months ago
by
supercharge19
An artificially high TruthfulQA, ironically, makes LLMs less truthful.
8
#1 opened 2 months ago
by
Phil337
using quants with pipeline
#1 opened 3 months ago
by
supercharge19
Will quantized versions be available?
1
#8 opened 3 months ago
by
supercharge19
SFT or DPO for function calling and agenting
#4 opened 3 months ago
by
supercharge19
Is instruction or chat model coming?
#4 opened 3 months ago
by
supercharge19
What is a stage and will quants be available?
2
#4 opened 3 months ago
by
supercharge19
Come on try quantizing this
#4 opened 3 months ago
by
supercharge19
what is new?
1
#2 opened 3 months ago
by
supercharge19
How to run changed format?
3
#2 opened 3 months ago
by
supercharge19
GGUF versions?
7
#1 opened 3 months ago
by
supercharge19
Why different sizes for same quants?
6
#1 opened 3 months ago
by
supercharge19
Thank you
#1 opened 3 months ago
by
supercharge19
How to quantize these models?
3
#30 opened 3 months ago
by
supercharge19
Comparison with v2.5 instruct?
3
#2 opened 3 months ago
by
supercharge19
How to quantize these models?
8
#2 opened 3 months ago
by
supercharge19
What is the difference between this and eagle?
1
#2 opened 3 months ago
by
supercharge19
why two versions in a day?
1
#1 opened 3 months ago
by
supercharge19
🚩 Report : Ethical issue(s)
2
#74 opened 10 months ago
by
liuaiyao
Might consider attribution
106
#10 opened 3 months ago
by
arthurmensch
Possible future contamination problem
3
#7 opened 3 months ago
by
supercharge19
How to use it for different classes?
1
#27 opened 3 months ago
by
supercharge19
Help with prompt for this (and probably all other models) model, please.
2
#1 opened 3 months ago
by
supercharge19
So is it instruct or base model?
4
#1 opened 3 months ago
by
supercharge19
Full Model?
9
#13 opened 3 months ago
by
LoadingALIAS
What is DS, CL, S-DS and S-CL
3
#10 opened 3 months ago
by
supercharge19
License
7
#4 opened 3 months ago
by
mrfakename
Any quantization possible?
1
#18 opened 3 months ago
by
supercharge19
doesn't work with llama.cpp
2
#1 opened 3 months ago
by
vasilee
Python bindings not working
#5 opened 3 months ago
by
supercharge19
Please add function calling code/example from llama cpp python
1
#1 opened 4 months ago
by
supercharge19
Instruct mode responses seems ok, but Chat-instruct responses are not relevant.
2
#1 opened 4 months ago
by
robert1968
quantized version?
#3 opened 4 months ago
by
supercharge19
How to quantize
3
#10 opened 4 months ago
by
supercharge19
Which CheckPoint Model is this based on? 1T tokens to 3T tokens training data
2
#1 opened 4 months ago
by
jasonden
Name too christian
1
#1 opened 4 months ago
by
supercharge19