[whatmodel.sh](../../README.md) / Falcon 7B Instruct
# Falcon 7B Instruct
TII · released 2023-04 · 7.22B params · apache-2.0
**score 25.6** ±2.2 · #684 of 732 · from 11 benchmark results across 7 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Falcon 7B Instruct among every model on that task or test, worst to best.
task vs the others rank counts tests
chat & writing ▁▁●▃▆▄▆█▆▇▇▃ #395 of 402* 42% 2
chat & writing · #395 of 402 · 42% of its score
Beats [Phi-2 (2.7B)](phi-2.md), [ChatGLM2-6B](chatglm2-6b.md), [Qwen1.5 1.8B Chat](qwen1-5-1-8b-chat.md) here, though they rank higher overall. Trails [Gemma 3 270M IT](gemma-3-270m.md), [SmolLM2 360M Instruct](smollm2-360m.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
AlpacaEval 2 length-controlled 4.0% ▁●▂▃▆▆█▆▄▄▁▁ beats 3/122 medium · avg 35% 28% sharp · too hard for it
AlpacaEval 1.0 win rate (vs text-… 45.7% ▂▂·●·▇▃▆█▆▄▂ beats 6/45 very easy · avg 94% 14% sharp · at its level · noisy
other tests (9)
test result vs the others in its run difficulty counts why
Natural Questions (closed book) 13.6% ▁·▁·▁▂▁▂·●▂█ beats 11/28 medium · avg 48% 28% typical
Vectara hallucination leaderboard… 70.1% ●▁▂▃▄▄█▆▃▂▂▁ beats 0/140 very easy · avg 94% 8.7% blunt · noisy
Conic10K mathQA (zero-shot CoT) 0.0% ●··▄·······▄ beats 0/3 very hard · avg 10% 5.4% too hard for it
TriviaQA 39.2% ▁··▁█▃▃▁●▅▄█ beats 14/28 easy · avg 73% 5.2% at its level · noisy
This is not a Dataset (negation),… 50.1% ●▁····▃▁·▁·▁ beats 1/14 hard · avg 66% 2.4% sharp · too hard for it · noisy
JEEBench total score (zero-shot, … 9.8% ●··█····█··█ beats 0/3 very hard · avg 20% 2.4% too hard for it
TruthfulQA MC2 44.1% ▅▃█●▁▂▅···▁▁ beats 17/31 medium · avg 57% 1.4% blunt · at its level
HaluEval QA hallucination detecti… 29.7% ●▁▁·▁▁·▁▁▁▁▁ beats 1/27 hard · avg 66% 1.4% too hard for it · noisy
FANToM BeliefQ [Choice], short co… 43.9% ●·····▂····▁ beats 4/11 very hard · avg 55% 0.3% sharp · too hard for it · noisy
source results share of the score
AlpacaEval 1.0 and 2.0 leaderboards 2 42% ▇▇▇▇▇
Hallucinations Leaderboard results 4 36% ▇▇▇▇
Vectara hallucination leaderboard 1 9% ▇
Conic10K: A Challenging Math Problem Un… 1 5% ▇
This is not a Dataset: A Large Negation… 1 2% ▇
Have LLMs Advanced Enough? A Challengin… 1 2% ▇
FANToM: A Benchmark for Stress-testing … 1 0% ▇
prior: pull toward the middle 2% ▇
weighted average = 25.6
Licences: AlpacaEval 1.0 and 2.0 leaderboards ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · Hallucinations Leaderboard results ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · Vectara hallucination leaderboard ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · Conic10K: A Challenging Math Problem Understanding and Reasoning Dataset ([CC BY 4.0](../../licenses/acl-2023-findings-emnlp-427-CC-BY-4-0.txt)) · This is not a Dataset: A Large Negation Benchmark to Challenge Large Language Models ([CC BY 4.0](../../licenses/acl-2023-emnlp-main-531-CC-BY-4-0.txt)) · Have LLMs Advanced Enough? A Challenging Problem Solving Benchmark For Large Language Models ([CC BY 4.0](../../licenses/acl-2023-emnlp-main-468-CC-BY-4-0.txt)) · FANToM: A Benchmark for Stress-testing Machine Theory of Mind in Interactions ([CC BY 4.0](../../licenses/acl-2023-emnlp-main-890-CC-BY-4-0.txt))
## How to run it
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run Falcon 7B Instruct (Q4_K_M, needs ~4.8 GB, from [huggingface.co/QuantFactory](https://huggingface.co/QuantFactory/falcon-7b-instruct-GGUF/tree/173e88f971dece5cf253dfd7acad84bcf05298c0))
```sh
llama cli -hf QuantFactory/falcon-7b-instruct-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf QuantFactory/falcon-7b-instruct-GGUF:Q4_K_M
```
Other versions: Q8_0 (~7.4 GB) `llama cli -hf QuantFactory/falcon-7b-instruct-GGUF:Q8_0`
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run Falcon 7B Instruct (Q4_K_M, needs ~4.8 GB, from [huggingface.co/QuantFactory](https://huggingface.co/QuantFactory/falcon-7b-instruct-GGUF/tree/173e88f971dece5cf253dfd7acad84bcf05298c0))
```sh
lms get https://huggingface.co/QuantFactory/falcon-7b-instruct-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: Q8_0 (~7.4 GB) `lms get https://huggingface.co/QuantFactory/falcon-7b-instruct-GGUF@Q8_0 -y`
### Ollama
1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download)
```sh
# Linux
curl -fsSL https://ollama.com/install.sh | sh
```
2. Run Falcon 7B Instruct (Q4_0, needs ~4.2 GB, from [ollama.com](https://ollama.com/library/falcon/tags))
```sh
ollama run falcon:7b
```
Model repository and documentation: [huggingface.co/tiiuae](https://huggingface.co/tiiuae/falcon-7b-instruct)
[interactive version](https://whatmodel.sh/model/falcon-7b-instruct) · [JSON](https://whatmodel.sh/api?format=json&all=1)