Meta Llama 3 70B Fp8 By FriendliAI: Benchmarks, Features and Detailed Analysis. Insights on Meta Llama 3 70B Fp8.

8-bit Autotrain compatible Base model:finetune:meta-llama... Base model:meta-llama/meta-lla... En Facebook Instruct Llama Llama-3 Meta Pytorch Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/FriendliAI/Meta-Llama-3-70B-fp8

Meta Llama 3 70B Fp8 Benchmarks

LLME Score: 0.14749

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

Meta Llama 3 70B Fp8 (FriendliAI/Meta-Llama-3-70B-fp8)

🌟 Advertise your project 🚀

Meta Llama 3 70B Fp8 Parameters and Internals

Model Type

text-generation

Use Cases

Areas:

Commercial, Research

Applications:

Assistant-like chat, Natural language generation

Primary Use Cases:

Instruction-tuned models are optimized for dialogue.

Limitations:

Use in languages other than English., Use in prohibited ways by the Acceptable Use Policy and Llama 3 Community License., Models trained on specific datasets; may produce inaccurate or biased outputs.

Considerations:

Developers may fine-tune for languages beyond English per the Community License and Policy.

Additional Notes

Enhanced inference efficiency by quantizing to FP8.

Training Details

Data Sources:

A new mix of publicly available online data

Data Volume:

15T+ tokens

Methodology:

fine-tuning with supervised fine-tuning (SFT) and reinforcement learning from human feedback (RLHF)

Context Length:

8000

Hardware Used:

Meta's Research SuperCluster, H100-80GB GPUs

Model Architecture:

Auto-regressive language model with optimized transformer architecture

Safety Evaluation

Methodologies:

adversarial evaluations, red-teaming exercises

Findings:

residual risks expected, emphasis on mitigations for over-refusing prompts

Risk Categories:

CBRNE threats, Cyber security, Child safety

Responsible Ai Considerations

Mitigation Strategies:

Implemented series of safety tools such as Meta Llama Guard 2 and Code Shield

Input Output

Input Format:

text

Accepted Modalities:

text

Output Format:

text, code

Release Notes

Version:

Date:

April 18, 2024

Notes:

Meta developed and released the Meta Llama 3 family of large language models (LLMs).

LLM Name	Meta Llama 3 70B Fp8
Repository 🤗	https://huggingface.co/FriendliAI/Meta-Llama-3-70B-fp8
Model Name	Llama 3 70B
Model Creator	Meta Llama 3
Base Model(s)	Meta Llama 3 70B Instruct meta-llama/Meta-Llama-3-70B-Instruct
Model Size	70b
Required VRAM	72.7 GB
Updated	2025-09-23
Maintainer	FriendliAI
Model Type	llama
Instruction-Based	Yes
Model Files	9.9 GB: 1-of-8 9.9 GB: 2-of-8 9.8 GB: 3-of-8 9.9 GB: 4-of-8 9.8 GB: 5-of-8 9.9 GB: 6-of-8 9.8 GB: 7-of-8 3.7 GB: 8-of-8
Supported Languages	en
Model Architecture	LlamaForCausalLM
License	other
Context Length	8192
Model Max Length	8192
Transformers Version	4.40.0
Tokenizer Class	PreTrainedTokenizerFast
Padding Token	<\|end_of_text\|>
Vocabulary Size	128256
Torch Data Type	bfloat16

Quantized Models of the Meta Llama 3 70B Fp8

Model	Likes	Downloads	VRAM
Meta Llama 3 70B GPTQ	3	27	39 GB

Best Alternatives to Meta Llama 3 70B Fp8

Best Alternatives	Context / RAM	Downloads	Likes
... Chat 1048K Chinese Llama3 70B	1024K / 141.9 GB	9826	5
... Chat 1048K Chinese Llama3 70B	1024K / 141.9 GB	9069	5
... 3 70B Instruct Gradient 1048K	1024K / 141.9 GB	13	122
Llama3 Function Calling 1048K	1024K / 141.9 GB	6	1
...a 3 70B Instruct Gradient 524K	512K / 141.9 GB	10	23
...a 3 70B Instruct Gradient 262K	256K / 141.9 GB	114	56
...ama 3 70B Arimas Story RP V2.0	256K / 141.1 GB	26	3
...ama 3 70B Arimas Story RP V1.6	256K / 141.2 GB	13	0
...ama 3 70B Arimas Story RP V1.5	256K / 141.2 GB	46	3
Llama 3.1 70B Instruct	128K / 141.9 GB	888766	849

Note: green Score (e.g. "73.2") means that the model is better than FriendliAI/Meta-Llama-3-70B-fp8.

Rank the Meta Llama 3 70B Fp8 Capabilities

🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟

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

What open-source LLMs or SLMs are you in search of? 51545 in total.

Email us: info@extractum.io. Our Privacy Policy | Terms and Conditions | Suggest an improvement.

Our Social Media →

Original data from HuggingFace, OpenCompass and various public git repos.

Release v20241124

Support LLM Explorer