What are the hardware requirements for Bagel DPO 20B V04 Llama?

Bagel DPO 20B V04 Llama requires approximately 39.6 GB of VRAM and supports a context window of 32K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Bagel DPO 20B V04 Llama and how large is it?

Bagel DPO 20B V04 Llama is developed by jondurbin, a model with 20b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Bagel DPO 20B V04 Llama?

Bagel DPO 20B V04 Llama is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Bagel DPO 20B V04 Llama by jondurbin — VRAM 39.6GB, 32K context

Name: Bagel DPO 20B V04 Llama
Author: jondurbin

Bagel DPO 20B V04 Llama is an open-source language model by jondurbin. Features: 20b LLM, VRAM: 39.6GB, Context: 32K, License: other, Instruction-Based, LLM Explorer Score: 0.11.

Base model:finetune:internlm/i... Base model:internlm/internlm2-... Conversational Dataset:ai2 arc Dataset:allenai/ultrafeedback ... Dataset:argilla/distilabel-int... Dataset:b-mc2/sql-create-conte... Dataset:bluemoon-fandom-1-1-rp... Dataset:boolq Dataset:cakiki/rosetta-code Dataset:camel-ai/biology Dataset:camel-ai/chemistry Dataset:camel-ai/math Dataset:camel-ai/physics Dataset:codeparrot/apps Dataset:facebook/belebele Dataset:glaiveai/glaive-functi... Dataset:grimulkan/limarp-augme... Dataset:jondurbin/airoboros-3.... Dataset:jondurbin/cinematika-v... Dataset:jondurbin/contextual-d... Dataset:jondurbin/gutenberg-dp... Dataset:jondurbin/py-dpo-v0.1 Dataset:jondurbin/truthy-dpo-v... Dataset:kingbri/pippa-sharegpt Dataset:ldjnr/capybara Dataset:lmsys/lmsys-chat-1m Dataset:mattpscott/airoboros-s... Dataset:migtissera/synthia-v1.... Dataset:muennighoff/natural-in... Dataset:open-orca/slimorca Dataset:openbookqa Dataset:parisneo/lollms aware ... Dataset:piqa Dataset:ropes Dataset:squad v2 Dataset:tiger-lab/mathinstruct Dataset:unalignment/toxic-dpo-... Dataset:vezora/tested-22k-pyth... Dataset:whiterabbitneo/wrn-cha... Dataset:whiterabbitneo/wrn-cha... Dataset:winogrande Dataset:wizardlm/wizardlm evol... Endpoints compatible Instruct Llama Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/jondurbin/bagel-dpo-20b-v04-llama

Bagel DPO 20B V04 Llama Parameters and Internals

Training Details

Data Sources:

ai2_arc, airoboros, apps, belebele, bluemoon, boolq, camel-ai biology, camel-ai chemistry, camel-ai math, camel-ai physics, capybara, cinematika, emobank, evol-instruct, glaive-function-calling-v2, gutenberg, limarp-augmented, lmsys_chat_1m, lollms, mathinstruct, natural_instructions, openbookqa, pippa, piqa, python_alpaca, ropes, rosetta_code, slimorca, sql-create-context, squad_v2, airoboros-summarization, synthia, whiterabbitneo chapter 1, whiterabbitneo chapter 2, winogrande

LLM Name	Bagel DPO 20B V04 Llama
Repository 🤗	https://huggingface.co/jondurbin/bagel-dpo-20b-v04-llama
Base Model(s)	Internlm2 20B internlm/internlm2-20b
Model Size	20b
Required VRAM	39.6 GB
Updated	2026-07-31
Maintainer	jondurbin
Model Type	llama
Instruction-Based	Yes
Model Files	9.9 GB: 1-of-4 9.9 GB: 2-of-4 9.9 GB: 3-of-4 9.9 GB: 4-of-4
Model Architecture	LlamaForCausalLM
License	other
Context Length	32768
Model Max Length	32768
Transformers Version	4.37.1
Is Biased	0
Tokenizer Class	InternLM2Tokenizer
Padding Token	</s>
Vocabulary Size	92544
Torch Data Type	bfloat16

Rank the Bagel DPO 20B V04 Llama capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Bagel DPO 20B V04 Llama

Best Alternatives	Context / RAM	Downloads	Likes
EAGLE3 GPT Oss 20B	128K / 0.7 GB	507	0
Bagel 20B V04 Llama	32K / 39.6 GB	6	7
...lm2 1.7B Distilled GPT Oss 20B	8K / 0 GB	35	3
...runed Base PostPruneRetraining	8K / 0 GB	5	0
...lerLM 20M Instruct Pruned Sft5	8K / 0.1 GB	12	0
Rose 20B	4K / 39.9 GB	114	46
...I Llama2 Ko En Instruct 20B V1	2K / 40.4 GB	5	0
Rose 20B GPTQ	4K / 10.5 GB	4	3
Rose 20B AWQ	4K / 10.9 GB	1	1

Note: green Score (e.g. "73.2") means that the model is better than jondurbin/bagel-dpo-20b-v04-llama.