LLM EXPLORER 61,652 MODELS INDEXED

Stella En 1.5B V5 by dunzhang

By dunzhang · 581890 downloads
HF Repo ↗ Copy URL

Stella En 1.5B V5 is an open-source language model by dunzhang. Features: 1.5b LLM, VRAM: 6.2GB, Context: 128K, License: mit, LLM Explorer Score: 0.23.

  Arxiv:2205.13147   Autotrain compatible   Custom code   Endpoints compatible   Model-index   Mteb   Pytorch   Qwen2   Region:us   Safetensors   Sentence-similarity   Sentence-transformers   Text-embeddings-inference
Model Card on HF 🤗: https://huggingface.co/dunzhang/stella_en_1.5B_v5 

Stella En 1.5B V5 Parameters and Internals

Model Type 
SentenceTransformer, Retrieval, Classification, STS, PairClassification, Clustering, Summarization
Additional Notes 
The model comes with a series of `2_Dense_{dims}` folders for versatile output dimensions with minor performance differences across them. It requires a prompt for the best results in specific tasks.
Training Details 
Methodology:
The models are trained based on `Alibaba-NLP/gte-large-en-v1.5` and `Alibaba-NLP/gte-Qwen2-1.5B-instruct` and use [MRL](https://arxiv.org/abs/2205.13147) methodology.
Context Length:
512
Input Output 
Input Format:
Given a web search query, retrieve relevant passages that answer the query: {query}
Output Format:
Vectors with specified dimension, capable of complex similarity and retrieval tasks.
Performance Tips:
Use 1024 dimensions for a balance between performance and computational cost.
LLM NameStella En 1.5B V5
Repository 🤗https://huggingface.co/dunzhang/stella_en_1.5B_v5 
Model Size1.5b
Required VRAM6.2 GB
Updated2025-01-23
Maintainerdunzhang
Model Typeqwen2
Model Files  6.2 GB   6.2 GB
Model ArchitectureQwen2ForCausalLM
Licensemit
Context Length131072
Model Max Length131072
Transformers Version4.42.3
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Vocabulary Size151646
Torch Data Typefloat32
Errorsreplace

Best Alternatives to Stella En 1.5B V5

Best Alternatives
Context / RAM
Downloads
Likes
Qwen2.5 1.5B 1.3M Stretched1280K / 3.1 GB180
ReaderLM V2500K / 3.1 GB35356804
Reader Lm 1.5B250K / 3.1 GB385608
DeepSeek R1 Distill Qwen 1.5B128K / 3.5 GB6000511545
Pulsar Coder V1128K / 3.1 GB2911
MLR Executor Qwen 1.5B128K / 3.5 GB5030
Axiom Python 1.5B128K / 3.1 GB3120
Ko R1 1.5B Preview128K / 7.1 GB89
...2 1.5B Lora Tensorboard Merged128K / 3.1 GB1271
Qwen2.5 1.5B128K / 3.1 GB863295203
Note: green Score (e.g. "73.2") means that the model is better than dunzhang/stella_en_1.5B_v5.