Downloads · 30 days
931
8% of all-time downloads
RichardErkhov/flammenai_-_flammen16-mistral-7B-gguf
flammenai_-_flammen16-mistral-7B-gguf is a machine learning model from RichardErkhov. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Downloads · 30 days
931
8% of all-time downloads
All-time downloads
12.3K
Public
Repo size
82.2 GB
Likes
0
Public
Click a slice to open those files.
.gguf95.2 GB · 100%
From the Hugging Face model README
Quantization made by Richard Erkhov.
flammen16-mistral-7B - GGUF
| Name | Quant method | Size |
|---|---|---|
| flammen16-mistral-7B.Q2_K.gguf | Q2_K | 2.53GB |
| flammen16-mistral-7B.IQ3_XS.gguf | IQ3_XS | 2.81GB |
| flammen16-mistral-7B.IQ3_S.gguf | IQ3_S | 2.96GB |
| flammen16-mistral-7B.Q3_K_S.gguf | Q3_K_S | 2.95GB |
| flammen16-mistral-7B.IQ3_M.gguf | IQ3_M | 3.06GB |
| flammen16-mistral-7B.Q3_K.gguf | Q3_K | 3.28GB |
| flammen16-mistral-7B.Q3_K_M.gguf | Q3_K_M | 3.28GB |
| flammen16-mistral-7B.Q3_K_L.gguf | Q3_K_L | 3.56GB |
| flammen16-mistral-7B.IQ4_XS.gguf | IQ4_XS | 3.67GB |
| flammen16-mistral-7B.Q4_0.gguf | Q4_0 | 3.83GB |
| flammen16-mistral-7B.IQ4_NL.gguf | IQ4_NL | 3.87GB |
| flammen16-mistral-7B.Q4_K_S.gguf | Q4_K_S | 3.86GB |
| flammen16-mistral-7B.Q4_K.gguf | Q4_K | 4.07GB |
| flammen16-mistral-7B.Q4_K_M.gguf | Q4_K_M | 4.07GB |
| flammen16-mistral-7B.Q4_1.gguf | Q4_1 | 4.24GB |
| flammen16-mistral-7B.Q5_0.gguf | Q5_0 | 4.65GB |
| flammen16-mistral-7B.Q5_K_S.gguf | Q5_K_S | 4.65GB |
| flammen16-mistral-7B.Q5_K.gguf | Q5_K | 4.78GB |
| flammen16-mistral-7B.Q5_K_M.gguf | Q5_K_M | 4.78GB |
| flammen16-mistral-7B.Q5_1.gguf | Q5_1 | 5.07GB |
| flammen16-mistral-7B.Q6_K.gguf | Q6_K | 5.53GB |
| flammen16-mistral-7B.Q8_0.gguf | Q8_0 | 7.17GB |
library_name: transformers license: apache-2.0 base_model:

A Mistral 7B LLM built from merging pretrained models and finetuning on Jon Durbin's Truthy DPO set. Flammen specializes in exceptional character roleplay, creative writing, and general intelligence
Finetuned using an A100 on Google Colab. 🙏
Fine-tune a Mistral-7b model with Direct Preference Optimization - Maxime Labonne
LoRA, model, and training settings:
# LoRA configuration
peft_config = LoraConfig(
r=16,
lora_alpha=16,
lora_dropout=0.05,
bias="none",
task_type="CAUSAL_LM",
target_modules=['k_proj', 'gate_proj', 'v_proj', 'up_proj', 'q_proj', 'o_proj', 'down_proj']
)
# Model to fine-tune
model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype=torch.bfloat16,
load_in_4bit=True
)
model.config.use_cache = False
# Reference model
ref_model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype=torch.bfloat16,
load_in_4bit=True
)
# Training arguments
training_args = TrainingArguments(
per_device_train_batch_size=2,
gradient_accumulation_steps=2,
gradient_checkpointing=True,
learning_rate=2e-5,
lr_scheduler_type="cosine",
max_steps=420,
save_strategy="no",
logging_steps=1,
output_dir=new_model,
optim="paged_adamw_32bit",
warmup_steps=100,
bf16=True,
report_to="wandb",
)
# Create DPO trainer
dpo_trainer = DPOTrainer(
model,
ref_model,
args=training_args,
train_dataset=dataset,
tokenizer=tokenizer,
peft_config=peft_config,
beta=0.1,
max_prompt_length=1024,
max_length=1536,
force_use_ref_model=True
)
# Fine-tune model with DPO
dpo_trainer.train()