hchung1017/linear-merge

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:10.7BQuant:FP8Ctx Length:4kLicense:mitArchitecture:Transformer Open Weights Warm

hchung1017/linear-merge is a 10.7 billion parameter language model developed by hchung1017. This model is created by linearly merging three base models: LDCC/LDCC-SOLAR-10.7B, hyeogi/SOLAR-10.7B-dpo-v1, and megastudyedu/M-SOLAR-10.7B-v1.3. Its primary characteristic is its derivation from a linear combination of existing SOLAR-10.7B variants, aiming to consolidate their strengths.

Loading preview...

Model Overview

hchung1017/linear-merge is a 10.7 billion parameter language model developed by hchung1017. This model is distinct due to its construction method, which involves a linear merge of three pre-existing SOLAR-10.7B based models. The base models integrated are:

  • LDCC/LDCC-SOLAR-10.7B
  • hyeogi/SOLAR-10.7B-dpo-v1
  • megastudyedu/M-SOLAR-10.7B-v1.3

Key Characteristics

  • Merged Architecture: Combines the characteristics and potential strengths of multiple SOLAR-10.7B variants.
  • Parameter Count: Operates with 10.7 billion parameters, placing it in the medium-sized LLM category.

Potential Use Cases

This model is suitable for applications that could benefit from a consolidated model derived from several specialized or fine-tuned versions of the SOLAR-10.7B architecture. Developers might consider this model for tasks where the individual strengths of the merged components are desired in a single entity, potentially offering a balanced performance across various benchmarks or specific domains that the base models excel in.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p