Llama 3 8B instruction fine-tune
Captured configuration records LlamaForCausalLM with about 8.03B parameters, and the model card describes a fine-tune based on Meta-Llama-3-8b-Instruct.
Open Source Model Profile · OwenArli
ArliAI-Llama-3-8B-Dolfin-v0.5
ArliAI-Llama-3-8B-Dolfin-v0.5 is an 8.03B-parameter Llama fine-tune from OwenArli. The model card lists Meta-Llama-3-8b-Instruct with Dolphin and WizardLM instruction data.
- Publisher
- OwenArli
- Task
- text-generation
- Model type
- llama
- License
- llama3
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
ArliAI-Llama-3-8B-Dolfin-v0.5 is published by OwenArli as a Llama text-generation model. The captured configuration identifies LlamaForCausalLM with model type llama, and Safetensors metadata reports 8,030,261,248 parameters. According to the model card, it fine-tunes Meta-Llama-3-8b-Instruct on improved Dolphin and WizardLM data.
Recorded capabilities
Dolphin and WizardLM instruction focus
According to the model card, the fine-tune uses an improved Dolphin and WizardLM dataset intended to improve instruction following and reduce refusals.
Documented QLoRA training setup
According to the model card, training took about 2 days on 2xRTX 3090 with 4-bit loading and QLoRA 64-rank 128-alpha for about 2% trainable weights.
Documented sequence and format notes
According to the model card, training used 2048 sequence length against an 8192 base context, and the card introduces an instruct format section.
Use cases in the source record
- Conversational instruction-following workflows using the documented Dolphin and WizardLM fine-tune setup.
- Context-length-sensitive testing that treats the card's 8192-context statement as an unverified publisher claim.
Limitations and unknowns
- No evaluation results were extracted from this record.
- The card's statement that the model still performs the same 8192 context is a publisher testing claim and was not independently verified.
- Dataset quality and training outcomes are publisher claims from the model card and were not independently verified.
- Provider state is historical snapshot data and should be refreshed before being presented as current.
Source and provenance
Source: OwenArli/ArliAI-Llama-3-8B-Dolfin-v0.5
Captured: Unknown. Processed: 2026-09-07T19:34:35.495422+00:00.
Based on Meta-Llama-3-8b-Instruct, and is governed by Meta Llama 3 License agreement: https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct This is a fine tune using an improved Dolphin and WizardLM dataset intended to make the model follow instructions better and refuse less. OpenLLM Benchmark: Training: 2048 sequence length since the dataset has an average length of under 1000 tokens, while the base model is 8192 sequence length. From testing it still performs the same 8192 context just fine. Training duration is around 2 days on 2xRTX 3090, using 4-bit loading and Qlora 64-rank 128-alpha resulting in ~2% trainable weights. I…
F001F002F003F004F005F006F007F009F010F011F012F013F014