What are the hardware requirements for ChemLLM 20B Chat SFT?

ChemLLM 20B Chat SFT requires approximately 40.6 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 ChemLLM 20B Chat SFT and how large is it?

ChemLLM 20B Chat SFT is developed by AI4Chem, 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 ChemLLM 20B Chat SFT?

ChemLLM 20B Chat SFT is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

ChemLLM 20B Chat SFT by AI4Chem — VRAM 40.6GB, 32K context

Name: ChemLLM 20B Chat SFT
Author: AI4Chem

ChemLLM 20B Chat SFT is an open-source language model by AI4Chem. Features: 20b LLM, VRAM: 40.6GB, Context: 32K, License: apache-2.0, LLM Explorer Score: 0.12.

Arxiv:2402.06852 Chemistry Conversational Custom code En Feature-extraction Internlm2 Region:us Safetensors Sharded Tensorflow Zh

Model Card on HF 🤗: https://huggingface.co/AI4Chem/ChemLLM-20B-Chat-SFT

ChemLLM 20B Chat SFT Parameters and Internals

Model Type

text generation

Use Cases

Areas:

research, commercial applications

Primary Use Cases:

chemistry, molecule science

Additional Notes

LLM may generate incorrect answers, and users should proofread at their own risk.

Supported Languages

en (English), zh (Chinese)

Input Output

Input Format:

Prompts in text format

Accepted Modalities:

text

Output Format:

Generated text outputs

Performance Tips:

Use Dialogue Templates and System Prompts from Agent Chepybara for improved performance.

Release Notes

Version:

1.0

Date:

2024-01-18

Notes:

ChemLLM-7B-Chat ver 1.0 released and open-sourced.

Version:

1.5

Date:

2024-03-23

Notes:

ChemLLM-1.5 updated with a demo site and API reference.

LLM Name	ChemLLM 20B Chat SFT
Repository 🤗	https://huggingface.co/AI4Chem/ChemLLM-20B-Chat-SFT
Model Size	20b
Required VRAM	40.6 GB
Updated	2026-08-25
Maintainer	AI4Chem
Model Type	internlm2
Model Files	1.1 GB: 1-of-41 1.0 GB: 2-of-41 1.0 GB: 3-of-41 1.0 GB: 4-of-41 1.0 GB: 5-of-41 1.0 GB: 6-of-41 1.0 GB: 7-of-41 1.0 GB: 8-of-41 1.0 GB: 9-of-41 1.0 GB: 10-of-41 1.0 GB: 11-of-41 1.0 GB: 12-of-41 1.0 GB: 13-of-41 1.0 GB: 14-of-41 1.0 GB: 15-of-41 1.0 GB: 16-of-41 1.0 GB: 17-of-41 1.0 GB: 18-of-41 1.0 GB: 19-of-41 1.0 GB: 20-of-41 1.0 GB: 21-of-41 1.0 GB: 22-of-41 1.0 GB: 23-of-41 1.0 GB: 24-of-41 1.0 GB: 25-of-41 1.0 GB: 26-of-41 1.0 GB: 27-of-41 1.0 GB: 28-of-41 1.0 GB: 29-of-41 1.0 GB: 30-of-41 1.0 GB: 31-of-41 1.0 GB: 32-of-41 1.0 GB: 33-of-41 1.0 GB: 34-of-41 1.0 GB: 35-of-41 1.0 GB: 36-of-41 1.0 GB: 37-of-41 1.0 GB: 38-of-41 1.0 GB: 39-of-41 0.4 GB: 40-of-41 1.1 GB: 41-of-41
Supported Languages	en zh
Model Architecture	InternLM2ForCausalLM
License	apache-2.0
Context Length	32768
Model Max Length	32768
Transformers Version	4.40.0
Is Biased	0
Vocabulary Size	92544
Torch Data Type	bfloat16

Rank the ChemLLM 20B Chat SFT 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 ChemLLM 20B Chat SFT

Best Alternatives	Context / RAM	Downloads	Likes
Internlm2 5 20B	256K / 39.7 GB	158	17
Internlm2 5 20B Chat	32K / 39.7 GB	7642	96
Internlm2 Chat 20B ExPO	32K / 39.8 GB	15821	1
Internlm2 20B	32K / 39.7 GB	20503	59
Internlm2 Chat 20B	32K / 39.7 GB	20635	88
Internlm2 Base 20B	32K / 39.7 GB	15804	8
...ternlm2 5 20B Chat Abliterated	32K / 39.7 GB	16	4
ChemLLM 20B Chat DPO	32K / 40.6 GB	33	10
Internlm Chatbode 20B	32K / 39.8 GB	4	1
MathGenie InterLM 20B	32K / 39.8 GB	12	10

Note: green Score (e.g. "73.2") means that the model is better than AI4Chem/ChemLLM-20B-Chat-SFT.