LLM EXPLORER
 LLM Explorer: A Curated Large Language Model Directory and Analytics
Dark Theme
  LLM  List    LLM Hosting    LLM Leaderboards    Blog    Newsfeed    Advertise

Madlad400 3B Mt Ct2 Int8 by Heng666

 »  All LLMs  »  Heng666  »  Madlad400 3B Mt Ct2 Int8   URL Share it on

  16   
  Arxiv:2309.04662   Aa   Abt   Ace   Acf   Ach   Ada   Adh   Ady   Af   Agr   Ahk   Ak   Akb   Alt   Alz   Am   Amu   Ang   Ann   Ape   Ar   Arn   As   Av   Awa   Ay   Az   Ba   Ban   Bas   Bbc   Bci   Be   Ber   Bew   Bg   Bgp   Bgz   Bho   Bi   Bik   Bim   Bjj   Bm   Bn   Bo   Bqc   Br   Bru   Brx   Bs   Bts   Btx   Bua   Bum   Bus   Bzj   Ca   Cab   Cac   Cak   Cbk   Cce   Ce   Ceb   Cfm   Ch   Chk   Chm   Chr   Cjk   Ckb   Cnh   Cni   Co   Crh   Crs   Cs   Ctu   Cuk   Cv   Cy   Da   Dataset:allenai/madlad-400   De   Din   Dje   Djk   Dov   Dtp   Dv   Dwr   Dyu   Dz   Ee   El   Emp   En   Endpoints compatible   Enq   Eo   Es   Et   Eu   Fa   Ff   Ffm   Fi   Fil   Fip   Fj   Fo   Fon   Fr   Frp   Fy   Ga   Gag   Gbm   Gd   Gil   Gl   Gn   Gof   Gom   Gor   Grc   Gsw   Gu   Gub   Guc   Guh   Gui   Gv   Gvl   Gym   Gyn   Ha   Haw   He   Hi   Hif   Hil   Hmn   Hne   Ho   Hr   Ht   Hu   Hui   Hus   Hvn   Hy   Iba   Ibb   Id   Ify   Ig   Ilo   Inb   Is   Iso   It   Iu   Ium   Izz   Ja   Jac   Jam   Jax   Jiv   Jv   Jvn   Ka   Kaa   Kac   Kbd   Kbp   Kek   Kg   Kha   Kj   Kjg   Kjh   Kk   Kl   Km   Kmb   Kn   Knj   Ko   Koi   Kos   Krc   Kri   Krj   Ks   Ksd   Ksw   Ktu   Ku   Kum   Kv   Kw   Kwi   Ky   La   Laj   Lb   Lg   Lhu   Ln   Lo   Lrc   Lt   Ltg   Lu   Lus   Luz   Lv   Mad   Mag   Mai   Mak   Mam   Mas   Maz   Mbt   Mdf   Mel   Meo   Meu   Mey   Mfe   Mg   Mgh   Mh   Mi   Min   Miq   Mk   Mkn   Ml   Mn   Mni   Mnw   Mps   Mqy   Mr   Mrj   Mrw   Ms   Msb   Msi   Msm   Mt   Mtr   Multilingual   My   Myv   Ne   New   Ngu   Nhe   Nia   Nij   Niq   Nl   Nnb   No   Noa   Noe   Nog   Nr   Nso   Nut   Nv   Ny   Nyu   Nzi   Oc   Oj   Om   Or   Os   Otq   Pa   Pag   Pam   Pap   Pau   Pck   Pis   Pl   Pon   Ppk   Prk   Ps   Pt   Qu   Qub   Quc   Quf   Quh   Qup   Quy   Qvc   Qvi   Qvz   Qxr   Raj   Rcf   Region:us   Rki   Rm   Rmc   Rn   Ro   Rom   Ru   Rw   Rwo   Sa   Sah   Sd   Sda   Se   Seh   Sg   Shn   Shp   Si   Sja   Sjp   Sk   Skr   Sl   Sm   Smt   Sn   So   Spp   Sq   Sr   Srm   Srn   Srr   Ss   St   Stq   Su   Sus   Suz   Sv   Sw   Sxn   Syl   Syr   Ta   Tab   Taj   Tbz   Tca   Tcy   Tdx   Te   Teo   Tet   Tg   Th   Ti   Tiv   Tk   Tks   Tlh   Tll   Tn   To   Toj   Tr   Translation   Trp   Ts   Tsc   Tsg   Tt   Tuc   Tvl   Twu   Tyv   Tyz   Tzh   Tzj   Tzo   Ubu   Udm   Ug   Uk   Ur   Uz   Ve   Vec   Vi   Wa   Wal   War   Wo   Xal   Xh   Xmm   Yap   Yi   Yo   Yua   Yue   Za   Zap   Zh   Zne   Zu   Zxx   Zza
Model Card on HF 🤗: https://huggingface.co/Heng666/madlad400-3b-mt-ct2-int8 

Madlad400 3B Mt Ct2 Int8 Benchmarks

LLME Score: 0.17799
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Madlad400 3B Mt Ct2 Int8 (Heng666/madlad400-3b-mt-ct2-int8)

Madlad400 3B Mt Ct2 Int8 Parameters and Internals

Model Type 
Language model
Use Cases 
Areas:
Research, Multilingual NLP tasks
Applications:
Machine Translation
Primary Use Cases:
Multilingual machine translation
Limitations:
Not domain-specific. Assessed only for research purposes.
Considerations:
Researchers must be cautious about biases and contextual uses.
Additional Notes 
The model is optimized for CTranslate2 and not initially built by the original authors.
Supported Languages 
multilingual (>400), primary (en), others (>400)
Training Details 
Data Sources:
MADLAD-400, publicly available data
Data Volume:
1 trillion tokens
Methodology:
Trained on MADLAD-400 using T5 architecture.
Model Architecture:
T5 architecture
Safety Evaluation 
Risk Categories:
bias, misinformation
Ethical Considerations:
Issues due to web-crawled datasets, potential toxic output, dual-use risks.
Responsible Ai Considerations 
Fairness:
Focus on languages underrepresented in corpora, addressing fairness by supporting 400+ languages.
Accountability:
Research community accountability.
Mitigation Strategies:
Preprocessing of training data to reduce biases.
Input Output 
Input Format:
<2xx> prepends input to specify target language.
Accepted Modalities:
text
Output Format:
Text output translation.
LLM NameMadlad400 3B Mt Ct2 Int8
Repository 🤗https://huggingface.co/Heng666/madlad400-3b-mt-ct2-int8 
Model Size3b
Required VRAM3 GB
Updated2025-02-22
MaintainerHeng666
Model Files  3.0 GB
Supported Languagesen ru es fr de it pt pl nl vi tr sv id ro cs zh hu ja th fi fa uk da el no bg sk ko ar lt ca sl he et lv hi sq ms az sr ta hr kk is ml mr te af gl be mk eu bn ka mn bs uz ur sw ne kn gu si cy eo la hy ky tg ga mt my km tt so ku ps pa rw lo ha dv fy lb mg gd am ug ht sd jv mi tk yi ba fo or xh su kl ny sm sn co zu ig yo st as oc cv br rm sa bo om se ce os lg ti ts ee av to tn fj ak dz ln gn wa sg lu ay qu za ve nv kv iu ho kw gv kj ss wo bm kg ch mh bi nr rn oj ks ff aa
Model ArchitectureAutoModel
Licenseapache-2.0

Best Alternatives to Madlad400 3B Mt Ct2 Int8

Best Alternatives
Context / RAM
Downloads
Likes
Falcon 3 3B Rp Lora Model0K / 0.1 GB02
Phi 4 Sky Lora0K / 0.3 GB02
Qwen 2.5 3B Lora Model0K / 0.1 GB01
Rocket 3B Llamafile0K /  GB50687
...230520 3147 520Gtokens Ctx40960K / 6 GB1050
Fastchat T5 3B V1.0 Ct2 Int80K / 2.9 GB22
RWKV 4 Raven 3B V11 Zh0K / 6 GB66
...edPajama INCITE Instruct 3B V10K / 5.5 GB93
...st RedPajama INCITE Chat 3B V10K / 5.5 GB44
Ct2fast Dolly V2 3B0K / 2.8 GB92
Note: green Score (e.g. "73.2") means that the model is better than Heng666/madlad400-3b-mt-ct2-int8.

Rank the Madlad400 3B Mt Ct2 Int8 Capabilities

🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟

Instruction Following and Task Automation  
Factuality and Completeness of Knowledge  
Censorship and Alignment  
Data Analysis and Insight Generation  
Text Generation  
Text Summarization and Feature Extraction  
Code Generation  
Multi-Language Support and Translation  

What open-source LLMs or SLMs are you in search of? 43470 in total.

Email us: info@extractum.io. Our Privacy Policy  | Terms and Conditions  | Suggest an improvement.
Our Social Media →  
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241227
  Support LLM Explorer