CodeFeedback tuning
According to the model card, the model was trained on the full 65k CodeFeedback dataset plus an additional 150k Code Feedback Filtered Instruction dataset.
Open Source Model Profile · rombodawg
Llama-3-8B-Instruct-Coder
Llama-3-8B-Instruct-Coder is an 8.03B-parameter Llama text-generation model from rombodawg. Its model card documents coding-focused fine-tuning on CodeFeedback datasets.
- Publisher
- rombodawg
- Task
- text-generation
- Model type
- llama
- License
- llama3
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
Llama-3-8B-Instruct-Coder is published by rombodawg as a Llama-based text-generation model. The captured configuration identifies LlamaForCausalLM and Safetensors metadata reports 8,030,261,248 parameters. According to the model card, it starts from Meta llama-3-8b-instruct via unsloth and is trained on combined CodeFeedback data.
Recorded capabilities
Qalore efficiency account
According to the model card, the Qalore method combines QLoRA with Galore techniques, fitting llama-3-8b in 14.5GB of VRAM and completing on an RTX A4000 16GB in 130 hours for under $20.
Use cases in the source record
- Code-focused text generation and instruction workflows using the documented CodeFeedback-trained behavior.
- Low-VRAM fine-tuning experiments following the card's Qalore method notes.
Limitations and unknowns
- No evaluation results were extracted from this record.
- No context-window value was extracted from this record.
- Training cost, time, and VRAM figures are publisher claims and were not independently verified.
- Provider state is historical snapshot data, not independently refreshed current availability.
Source and provenance
Source: rombodawg/Llama-3-8B-Instruct-Coder
Captured: Unknown. Processed: 2026-09-07T19:34:57.190676+00:00.
llama-3-8B-Instruct-Coder This model is llama-3-8b-instruct from Meta (uploaded by unsloth) trained on the full 65k Codefeedback dataset + the additional 150k Code Feedback Filtered Instruction dataset combined. You can find that dataset linked below. This AI model was trained with the new Qalore method developed by my good friend on Discord and fellow Replete-AI worker walmartbag. The Qalore method uses Qlora training along with the methods from Galore for additional reductions in VRAM allowing for llama-3-8b to be loaded on 14.5 GB of VRAM. This allowed this training to be completed on an RTX A4000 16GB in 130 hours for less than…
F001F002F003F004F005F006F007F009F010F011F012