geonho1
Mistral-7B-Instruct-v0.2-4b-r128-task1280
Available on FriendliAI
Dedicated EndpointsRun this model inference on single tenant GPU with unmatched speed and reliability at scale.
ContainerRun this model inference with full control and performance in your environment.
Model Details
Model Tree
Base
mistralai/Mistral-7B-Instruct-v0.2
Supported Functionality
Dedicated EndpointsContainer
README
License: apache-2.0
Source
- Base model:
mistralai/Mistral-7B-Instruct-v0.2
- Dataset:
Lots-of-LoRAs/task1280_ted_translation_pt_it
- Train split:
train
- Eval split:
valid
- Task ID:
1280
- Description:
ted translation pt it
LoRA
- Rank:
128
- Target modules:
q_proj, k_proj, v_proj
- LoRA alpha:
32
- LoRA dropout:
0.05
- Bias:
none
Training protocol
- Base model dtype:
4bit-nf4
- Quantization:
QLoRA 4bit NF4, double quantization enabled, bf16 compute
- Adapter trainable dtype:
float32
- Prompt format:
plain
- Loss: completion-only causal LM cross entropy
- Epochs:
5.0
- Best checkpoint metric:
eval_loss
- Learning rate:
0.0002
- Scheduler:
cosine
- Warmup ratio:
Files
adapter_model.safetensors: LoRA adapter weightsadapter_config.json: PEFT adapter configurationtask_manifest.json: source manifest row and resolved splitstraining_protocol.json: fixed protocol used for this run
0.03
Effective batch size: 16Optimizer: paged_adamw_32bit