LLM EXPLORER 61,249 MODELS INDEXED

Moss Moon 003 Sft Int4 Fix Autotune by DotIN13

By DotIN13 · 18 downloads

Moss Moon 003 Sft Int4 Fix Autotune is an open-source language model by DotIN13. Features: LLM, VRAM: 10.8GB, License: agpl-3.0, LLM Explorer Score: 0.07.

  Arxiv:2203.13474   Custom code   Dataset:fnlp/moss-002-sft-data   En   Moss   Pytorch   Region:us   Zh

Moss Moon 003 Sft Int4 Fix Autotune Parameters and Internals

Model Type 
Conversational language model
Use Cases 
Areas:
Research, Commercial applications
Applications:
Multi-turn dialogues, Plugin-augmented functions
Primary Use Cases:
Conversational AI, Language-based tasks
Limitations:
Can generate incorrect, misleading, or biased information.
Considerations:
Carefully check contents generated by the model before use.
Additional Notes 
Possible improvement plans involve reasoning, truthfulness, safety, multi-modality, and personalization.
Supported Languages 
en (English), zh (Chinese)
Training Details 
Data Sources:
The PILE, BigQuery, BigPython, Private Chinese corpus
Data Volume:
~700B tokens
Methodology:
Supervised Fine-tuning, Preference-aware training
Context Length:
2048
Training Time:
Not specified
Hardware Used:
2 NVIDIA 3090 GPUs with FP16 precision
Model Architecture:
Initialized with CodeGen and fine-tuned on multi-turn plugin-augmented conversational data.
Responsible Ai Considerations 
Fairness:
Not specified
Transparency:
Not specified
Accountability:
Unspecified
Mitigation Strategies:
Not specified
Input Output 
Input Format:
text input
Accepted Modalities:
text
Output Format:
text output
Performance Tips:
Use FP16 precision for enhanced performance on A100 or similar GPUs.
LLM NameMoss Moon 003 Sft Int4 Fix Autotune
Repository πŸ€—https://huggingface.co/DotIN13/moss-moon-003-sft-int4-fix-autotune 
Required VRAM10.8 GB
Updated2026-08-17
MaintainerDotIN13
Model Typemoss
Model Files  10.8 GB
Supported Languagesen zh
Model ArchitectureMossForCausalLM
Licenseagpl-3.0
Model Max Length2048
Transformers Version4.25.1
Tokenizer ClassGPT2Tokenizer
Beginning of Sentence Token<|endoftext|>
End of Sentence Token<eom>
Unk Token<|endoftext|>
Vocabulary Size107008
Torch Data Typefloat16
Activation Functiongelu_new
Errorsreplace

Best Alternatives to Moss Moon 003 Sft Int4 Fix Autotune

Best Alternatives
Context / RAM
Downloads
Likes
Moss Moon 003 Base0K / 33.5 GB247131
Moss Moon 003 Sft0K / 33.5 GB131129
Moss Moon 003 Base0K / 33.5 GB67132
Moss Moon 003 Sft0K / 33.5 GB21127
Moss Moon 003 Sft Plugin0K / 33.5 GB2672
Moss Moon 003 Sft Plugin0K / 33.5 GB769
Moss Moon 003 Sft Int40K / 10.8 GB4240
Moss Moon 003 Sft Int40K / 10.8 GB2941
Moss Moon 003 Sft Plugin Int40K / 10.8 GB2019
Moss Moon 003 Sft Int80K / 18.5 GB2814
Note: green Score (e.g. "73.2") means that the model is better than DotIN13/moss-moon-003-sft-int4-fix-autotune.