Downloads · 30 days
3
11% of all-time downloads
myyycroft/Gemma-4-E2B-AmbigQA-lora-8-short-form-prompt-member-4
Gemma-4-E2B-AmbigQA-lora-8-short-form-prompt-member-4 is a text generation model from myyycroft. Use it when you need the model to write or continue text. It is set up for peft.
Ensemble member 4 (seed 4078) from run gemma4e2blora8smalllrshortformprompt.
Downloads · 30 days
3
11% of all-time downloads
All-time downloads
28
Public
Repo size
86.4 MB
Likes
0
Public
Click a slice to open those files.
.safetensors52.7 MB · 51%
From the Hugging Face model README
Ensemble member 4 (seed 4078) from run gemma4_e2b_lora_8_small_lr_short_form_prompt.
Fine-tuned from google/gemma-4-E2B-it on AmbigQA.
Metrics below are computed on small fixed subsets, not full benchmarks.
Exact subset sizes: AmbigQA (128), IFEval (64), MMLU (228).
Reported values are from the final training step (step=1620, epoch=3).
sewon/ambig_qa (config_name=light)6e667596df70f17ba3c8e7be4b7361f6be8b60f8train, validation=validationdev_fraction=0.1, split_seed=1729FalseYou answer open-domain factoid questions with a short-form answer only.
Reply with just the answer string: typically a name, place, organization, title, date, number, or brief factual phrase (often about 1–4 words). Do not write a sentence, explanation, or preamble.
Examples of the expected style:
- Who painted The Persistence of Memory? → Salvador Dalí
- When was the first iPhone released in the United States? → June 29, 2007
- What is the capital of New Zealand? → Wellington
- Who wrote Pride and Prejudice? → Jane Austen
- What is the chemical symbol for gold? → Au
- Who directed the 1994 film Pulp Fiction? → Quentin Tarantino
Formatting rules:
- Output the answer alone on a single line.
- Do not restate the question, add reasoning, markdown, bullets, or quotation marks around the answer.
- Prefer a canonical short form over a descriptive sentence.
- Give one answer only; do not list alternatives or hedge.
- Use natural short forms for dates and names (e.g. June 29, 2007; Jane Austen).
Exact resolved settings used for this run (also attached as resolved_config.yaml):
model:
name: google/gemma-4-E2B-it
revision: main
dtype: bfloat16
enable_thinking: false
dataset:
name: sewon/ambig_qa
config_name: light
revision: 6e667596df70f17ba3c8e7be4b7361f6be8b60f8
train_split: train
validation_split: validation
dev_fraction: 0.1
split_seed: 1729
bootstrap: false
max_train_examples: null
max_dev_examples: null
system_prompt_set: true
adaptation:
method: lora
lora:
r: 8
alpha: 16
dropout: 0.0
target_modules: all-linear
bias: none
training:
num_train_epochs: 3.0
max_steps: -1
learning_rate: 5.0e-05
weight_decay: 0.01
warmup_ratio: 0.03
per_device_train_batch_size: 16
per_device_eval_batch_size: 32
gradient_accumulation_steps: 2
max_seq_length: 512
logging_steps: 10
save_every_steps: 150
save_total_limit: 3
gradient_checkpointing: false
bf16: true
tf32: true
max_grad_norm: 1.0
dataloader_num_workers: 2
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1.0e-08
ensemble:
size: 5
base_seed: 42
this_member_index: 4
this_member_seed: 4078
evaluation_subsets:
ambigqa:
subset_size: 128
seed: 1001
ifeval:
subset_size: 64
seed: 1002
mmlu:
subset_size: 228
seed: 1004
The metrics in this section use small fixed subsets, not full benchmarks: AmbigQA (128), IFEval (64), MMLU (228).
| member | seed | steps | epochs | train_loss | AmbigQA (128) acc | AmbigQA (128) AlignScore | IFEval (64) strict | MMLU (228) acc |
|---|---|---|---|---|---|---|---|---|
| 0 | 42 | 1620 | 3 | 1.7112 | 0.0859 | 0.1877 | 0.6094 | 0.5044 |
| 1 | 1051 | 1620 | 3 | 1.7032 | 0.0859 | 0.1849 | 0.6875 | 0.5746 |
| 2 | 2060 | 1620 | 3 | 1.7041 | 0.1172 | 0.2043 | 0.5625 | 0.5044 |
| 3 | 3069 | 1620 | 3 | 1.7005 | 0.1016 | 0.1818 | 0.5781 | 0.5307 |
| 4 (this repo) | 4078 | 1620 | 3 | 1.7063 | 0.1172 | 0.1925 | 0.6875 | 0.5658 |
members/member_004/final/resolved_config.yaml — full resolved training configensemble_metrics.png — train/eval curves for the whole ensemblerun_artifacts/ — ensemble-level manifests, status, and captured environmentrun_artifacts/members/member_004/ — this member's manifests, metadata, and statusrun_artifacts/members/member_004/predictions/ — per-dataset JSONL predictions from every intermediate evaluation step