synapsellm-7b-mistral-v0.4-preview2/README.md

language, license, library_name, tags, model-index

language	license	library_name	tags	model-index
en	apache-2.0	transformers	code	name results synapsellm-7b-mistral-v0.4-preview2 task dataset metrics source type name text-generation Text Generation name type config split args AI2 Reasoning Challenge (25-Shot) ai2_arc ARC-Challenge test num_few_shot 25 type value name acc_norm 52.99 normalized accuracy url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=WebraftAI/synapsellm-7b-mistral-v0.4-preview2 Open LLM Leaderboard task dataset metrics source type name text-generation Text Generation name type split args HellaSwag (10-Shot) hellaswag validation num_few_shot 10 type value name acc_norm 74.54 normalized accuracy url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=WebraftAI/synapsellm-7b-mistral-v0.4-preview2 Open LLM Leaderboard task dataset metrics source type name text-generation Text Generation name type config split args MMLU (5-Shot) cais/mmlu all test num_few_shot 5 type value name acc 54.6 accuracy url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=WebraftAI/synapsellm-7b-mistral-v0.4-preview2 Open LLM Leaderboard task dataset metrics source type name text-generation Text Generation name type config split args TruthfulQA (0-shot) truthful_qa multiple_choice validation num_few_shot 0 type value mc2 53.79 url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=WebraftAI/synapsellm-7b-mistral-v0.4-preview2 Open LLM Leaderboard task dataset metrics source type name text-generation Text Generation name type config split args Winogrande (5-shot) winogrande winogrande_xl validation num_few_shot 5 type value name acc 73.95 accuracy url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=WebraftAI/synapsellm-7b-mistral-v0.4-preview2 Open LLM Leaderboard task dataset metrics source type name text-generation Text Generation name type config split args GSM8k (5-shot) gsm8k main test num_few_shot 5 type value name acc 25.7 accuracy url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=WebraftAI/synapsellm-7b-mistral-v0.4-preview2 Open LLM Leaderboard

SynapseLLM:

SynapseLLM, a significant achievement by WebraftAI, represents a series of large language AI models designed to create robust, generalized, and decentralized information systems. This repository specifically houses the SynapseLLM finetuned version of Mistral. The finetuning process is conducted on a custom dataset, albeit limited in scope, focusing on code and normal question-answering scenarios. This adaptation showcases the model's versatility and applicability within specific domains, contributing to the broader landscape of AI advancements.

Model Details

SynapseLLM:

• Parameters: 7B
• Learning rate: 2e-4
• Adapter used: Qlora
• Precision: float16
• Batch size: 32
• Maximum gradient normal: 0.3
• Optimizer: paged_adamw_32bit
• Warmup Ratio: 0.03
• Step(s) (trained): 150
• Epoch(s) (trained): 1

Model Description

This is a 7b parameter, decoder only transformer based finetuned model on Chat Q/A and Code instructions. It's a preview finetune on Mistral 7B v0.1 on a sample dataset of 770k rows comprising of 361k Maths Instruct Q/A, 143k GPT-3.5 Q/A, 140k General Code, 63k Python code, and 54k General Q/A (Through GPT-4) [Each row contains one instruction and one response]. This is a full model merged and compiled with trained adapters, so you can easily load this through transformers library.

• Developed by: WebraftAI
• Funded by: Webraft Cloud
• Shared by: WebraftAI
• Model type: Decoder-only Transformer
• Language(s): English Only
• License: Apache 2.0
• Finetuned from model: Mistral-7b-v0.1

Prompt format:

This model follows the same prompt format as mistral instruct 7b v0.1 .The sample prompt is still given below:

<s>[INST] Hello, how are you? [/INST]

Example Code:

Here's an example code using transformers library provided by HF.

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("WebraftAI/synapsellm-7b-mistral-v0.4-preview2")
model = AutoModelForCausalLM.from_pretrained("WebraftAI/synapsellm-7b-mistral-v0.4-preview2")

prompt= "<s>[INST] Hello!  [/INST] "

device = "cuda"

model_inputs = tokenizer([prompt], return_tensors="pt").to(device)
model.to(device)

generated_ids = model.generate(**model_inputs, max_new_tokens=100, do_sample=True)
print(tokenizer.batch_decode(generated_ids)[0])

Model Bias:

This model has some bias areas, discussed below:

• Model might output factually incorrect information.
• Model does not follow system prompts.
• Model does not have any kind of memory, researchers can experiment feeding memory.
• Model is trained on different datas, so it can bias information or exclaim itself as gpt model.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	55.93
AI2 Reasoning Challenge (25-Shot)	52.99
HellaSwag (10-Shot)	74.54
MMLU (5-Shot)	54.60
TruthfulQA (0-shot)	53.79
Winogrande (5-shot)	73.95
GSM8k (5-shot)	25.70