metadata
license: other
tags:
- axolotl
- generated_from_trainer
- Mistral
- instruct
- finetune
- chatml
- gpt4
- synthetic data
- science
- physics
- chemistry
- biology
- math
base_model: mistralai/Mistral-7B-v0.1
datasets:
- allenai/ai2_arc
- camel-ai/physics
- camel-ai/chemistry
- camel-ai/biology
- camel-ai/math
- metaeval/reclor
- openbookqa
- mandyyyyii/scibench
- derek-thomas/ScienceQA
- TIGER-Lab/ScienceEval
- jondurbin/airoboros-3.2
- LDJnr/Capybara
- Cot-Alpaca-GPT4-From-OpenHermes-2.5
- STEM-AI-mtl/Electrical-engineering
- knowrohit07/saraswati-stem
- sablo/oasst2_curated
- glaiveai/glaive-code-assistant
- lmsys/lmsys-chat-1m
- TIGER-Lab/MathInstruct
- bigbio/med_qa
- meta-math/MetaMathQA-40K
- openbookqa
- piqa
- metaeval/reclor
- derek-thomas/ScienceQA
- scibench
- sciq
- Open-Orca/SlimOrca
- migtissera/Synthia-v1.3
- TIGER-Lab/ScienceEval
model-index:
- name: Einstein-v4-7B
results:
- task:
type: text-generation
name: Text Generation
dataset:
name: AI2 Reasoning Challenge (25-Shot)
type: ai2_arc
config: ARC-Challenge
split: test
args:
num_few_shot: 25
metrics:
- type: acc_norm
value: 64.68
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Weyaxi/Einstein-v4-7B
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: HellaSwag (10-Shot)
type: hellaswag
split: validation
args:
num_few_shot: 10
metrics:
- type: acc_norm
value: 83.75
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Weyaxi/Einstein-v4-7B
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MMLU (5-Shot)
type: cais/mmlu
config: all
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 62.31
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Weyaxi/Einstein-v4-7B
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: TruthfulQA (0-shot)
type: truthful_qa
config: multiple_choice
split: validation
args:
num_few_shot: 0
metrics:
- type: mc2
value: 55.15
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Weyaxi/Einstein-v4-7B
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: Winogrande (5-shot)
type: winogrande
config: winogrande_xl
split: validation
args:
num_few_shot: 5
metrics:
- type: acc
value: 76.24
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Weyaxi/Einstein-v4-7B
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: GSM8k (5-shot)
type: gsm8k
config: main
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 57.62
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Weyaxi/Einstein-v4-7B
name: Open LLM Leaderboard
🔬 Einstein-v4-7B
This model is a full fine-tuned version of mistralai/Mistral-7B-v0.1 on diverse datasets.
This model is finetuned using 7xRTX3090 + 1xRTXA6000 using axolotl.
This model's training was sponsored by sablo.ai.
See axolotl config
axolotl version: 0.4.0
base_model: mistralai/Mistral-7B-v0.1
model_type: MistralForCausalLM
tokenizer_type: LlamaTokenizer
is_mistral_derived_model: true
load_in_8bit: false
load_in_4bit: false
strict: false
chat_template: chatml
datasets:
- path: data/merged_all.json
ds_type: json
type: alpaca
conversation: chatml
- path: data/capybara_sharegpt.json
ds_type: json
type: sharegpt
conversation: chatml
- path: data/synthia-v1.3_sharegpt_12500.json
ds_type: json
type: sharegpt
conversation: chatml
- path: data/cot_alpaca_gpt4_extracted_openhermes_2.5_sharegpt.json
ds_type: json
type: sharegpt
conversation: chatml
- path: data/slimorca_dedup_filtered_95k_sharegpt.json
ds_type: json
type: sharegpt
conversation: chatml
- path: data/airoboros_3.2_without_contextual_slimorca_orca_sharegpt.json
ds_type: json
type: sharegpt
conversation: chatml
dataset_prepared_path: last_run_prepared
val_set_size: 0.005
output_dir: ./Einstein-v4-model
sequence_len: 8192
sample_packing: true
pad_to_sequence_len: true
eval_sample_packing: false
wandb_project: Einstein
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:
hub_model_id: Weyaxi/Einstein-v4-7B
save_safetensors: true
gradient_accumulation_steps: 4
micro_batch_size: 1
num_epochs: 1.5
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.000005
train_on_inputs: false
group_by_length: false
bf16: true
fp16: false
tf32: false
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
warmup_steps: 10
evals_per_epoch: 2 # changed
eval_table_size:
eval_table_max_new_tokens: 128
saves_per_epoch: 4
debug:
deepspeed: zero3_bf16.json
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
bos_token: "<s>"
eos_token: "<|im_end|>"
unk_token: "<unk>"
tokens:
- "<|im_start|>"
resume_from_checkpoint: Einstein-v4-model/checkpoint-521
💬 Prompt Template
You can use this prompt template while using the model:
ChatML
<|im_start|>system
{system}<|im_end|>
<|im_start|>user
{user}<|im_end|>
<|im_start|>assistant
{asistant}<|im_end|>
This prompt template is available as a chat template, which means you can format messages using the
tokenizer.apply_chat_template() method:
messages = [
{"role": "system", "content": "You are helpful AI asistant."},
{"role": "user", "content": "Hello!"}
]
gen_input = tokenizer.apply_chat_template(message, return_tensors="pt")
model.generate(**gen_input)
🔄 Quantizationed versions
Quantizationed versions of this model is available.
Exl2 @bartowski:
You can switch up branches in the repo to use the one you want
| Branch | Bits | lm_head bits | VRAM (4k) | VRAM (16k) | VRAM (32k) | Description |
|---|---|---|---|---|---|---|
| 8_0 | 8.0 | 8.0 | 8.4 GB | 9.8 GB | 11.8 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
| 6_5 | 6.5 | 8.0 | 7.2 GB | 8.6 GB | 10.6 GB | Very similar to 8.0, good tradeoff of size vs performance, recommended. |
| 5_0 | 5.0 | 6.0 | 6.0 GB | 7.4 GB | 9.4 GB | Slightly lower quality vs 6.5, but usable on 8GB cards. |
| 4_25 | 4.25 | 6.0 | 5.3 GB | 6.7 GB | 8.7 GB | GPTQ equivalent bits per weight, slightly higher quality. |
| 3_5 | 3.5 | 6.0 | 4.7 GB | 6.1 GB | 8.1 GB | Lower quality, only use if you have to. |
🎯 Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 66.62 |
| AI2 Reasoning Challenge (25-Shot) | 64.68 |
| HellaSwag (10-Shot) | 83.75 |
| MMLU (5-Shot) | 62.31 |
| TruthfulQA (0-shot) | 55.15 |
| Winogrande (5-shot) | 76.24 |
| GSM8k (5-shot) | 57.62 |
🤖 Additional information about training
This model is full fine-tuned for 1.5 epoch.
Total number of steps was 1562.
Loss graph
🤝 Acknowledgments
Thanks to sablo.ai for sponsoring this model.
Thanks to all the dataset authors mentioned in the datasets section.
Thanks to axolotl for making the repository I used to make this model.
Thanks to all open source AI community.
If you would like to support me: