metadata
library_name: llama.cpp
license: gemma
tags: []
widget:
- text: |
<start_of_turn>user
How does the brain work?<end_of_turn>
<start_of_turn>model
inference:
parameters:
max_new_tokens: 200
extra_gated_heading: Access Gemma on Hugging Face
extra_gated_prompt: >-
To access Gemma on Hugging Face, you’re required to review and agree to
Google’s usage license. To do this, please ensure you’re logged-in to Hugging
Face and click below. Requests are processed immediately.
extra_gated_button_content: Acknowledge license
Gemma Model Card
Model Page: Gemma
This model card corresponds to the 2B instruct version of the Gemma model in GGUF Format. The weights here are float32.
In llama.cpp, and other related tools such as Ollama and LM Studio, please make sure that you have these flags set correctly, especially
repeat-penalty. Georgi Gerganov (llama.cpp's author) shared his experience in https://huggingface.co/google/gemma-7b-it/discussions/38#65d7b14adb51f7c160769fa1.
You can also visit the model card of the 2B base model GGUF, 7B base model GGUF, and 7B instruct model GGUF.
Resources and Technical Documentation:
Terms of Use: Terms
Authors: Google
Model Information
Summary description and brief definition of inputs and outputs.
Description
Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models. They are text-to-text, decoder-only large language models, available in English, with open weights, pre-trained variants, and instruction-tuned variants. Gemma models are well-suited for a variety of text generation tasks, including question answering, summarization, and reasoning. Their relatively small size makes it possible to deploy them in environments with limited resources such as a laptop, desktop or your own cloud infrastructure, democratizing access to state of the art AI models and helping foster innovation for everyone.
Usage
Below we share some commands on how to get quickly started with running the model.
Running the model on a CPU
llama.cpp/build$ bin/main -m gemma-2b-it.gguf -n 256 -p "Write a Python function to find sum of two numbers." --repeat-penalty 1.1
Log start
main: build = 2249 (15499eb9)
main: built with cc (Debian 13.2.0-5) 13.2.0 for x86_64-linux-gnu
main: seed = 1708973831
llama_model_loader: loaded meta data with 19 key-value pairs and 164 tensors from gemma-2b-it.gguf (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv 0: general.architecture str = gemma
llama_model_loader: - kv 1: general.name str = gemma-2b-it
llama_model_loader: - kv 2: gemma.context_length u32 = 8192
llama_model_loader: - kv 3: gemma.block_count u32 = 18
llama_model_loader: - kv 4: gemma.embedding_length u32 = 2048
llama_model_loader: - kv 5: gemma.feed_forward_length u32 = 16384
llama_model_loader: - kv 6: gemma.attention.head_count u32 = 8
llama_model_loader: - kv 7: gemma.attention.head_count_kv u32 = 1
llama_model_loader: - kv 8: gemma.attention.key_length u32 = 256
llama_model_loader: - kv 9: gemma.attention.value_length u32 = 256
llama_model_loader: - kv 10: gemma.attention.layer_norm_rms_epsilon f32 = 0.000001
llama_model_loader: - kv 11: tokenizer.ggml.model str = llama
llama_model_loader: - kv 12: tokenizer.ggml.bos_token_id u32 = 2
llama_model_loader: - kv 13: tokenizer.ggml.eos_token_id u32 = 1
llama_model_loader: - kv 14: tokenizer.ggml.padding_token_id u32 = 0
llama_model_loader: - kv 15: tokenizer.ggml.unknown_token_id u32 = 3
llama_model_loader: - kv 16: tokenizer.ggml.tokens arr[str,256128] = ["<pad>", "<eos>", "<bos>", "<unk>", ...
llama_model_loader: - kv 17: tokenizer.ggml.scores arr[f32,256128] = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv 18: tokenizer.ggml.token_type arr[i32,256128] = [3, 3, 3, 2, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - type f32: 164 tensors
llm_load_vocab: mismatch in special tokens definition ( 544/256128 vs 388/256128 ).
llm_load_print_meta: format = GGUF V3 (latest)
llm_load_print_meta: arch = gemma
llm_load_print_meta: vocab type = SPM
llm_load_print_meta: n_vocab = 256128
llm_load_print_meta: n_merges = 0
llm_load_print_meta: n_ctx_train = 8192
llm_load_print_meta: n_embd = 2048
llm_load_print_meta: n_head = 8
llm_load_print_meta: n_head_kv = 1
llm_load_print_meta: n_layer = 18
llm_load_print_meta: n_rot = 256
llm_load_print_meta: n_embd_head_k = 256
llm_load_print_meta: n_embd_head_v = 256
llm_load_print_meta: n_gqa = 8
llm_load_print_meta: n_embd_k_gqa = 256
llm_load_print_meta: n_embd_v_gqa = 256
llm_load_print_meta: f_norm_eps = 0.0e+00
llm_load_print_meta: f_norm_rms_eps = 1.0e-06
llm_load_print_meta: f_clamp_kqv = 0.0e+00
llm_load_print_meta: f_max_alibi_bias = 0.0e+00
llm_load_print_meta: n_ff = 16384
llm_load_print_meta: n_expert = 0
llm_load_print_meta: n_expert_used = 0
llm_load_print_meta: rope scaling = linear
llm_load_print_meta: freq_base_train = 10000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_yarn_orig_ctx = 8192
llm_load_print_meta: rope_finetuned = unknown
llm_load_print_meta: model type = 2B
llm_load_print_meta: model ftype = all F32 (guessed)
llm_load_print_meta: model params = 2.51 B
llm_load_print_meta: model size = 9.34 GiB (32.00 BPW)
llm_load_print_meta: general.name = gemma-2b-it
llm_load_print_meta: BOS token = 2 '<bos>'
llm_load_print_meta: EOS token = 1 '<eos>'
llm_load_print_meta: UNK token = 3 '<unk>'
llm_load_print_meta: PAD token = 0 '<pad>'
llm_load_print_meta: LF token = 227 '<0x0A>'
llm_load_tensors: ggml ctx size = 0.06 MiB
llm_load_tensors: CPU buffer size = 9561.29 MiB
.............................................................
llama_new_context_with_model: n_ctx = 512
llama_new_context_with_model: freq_base = 10000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init: CPU KV buffer size = 9.00 MiB
llama_new_context_with_model: KV self size = 9.00 MiB, K (f16): 4.50 MiB, V (f16): 4.50 MiB
llama_new_context_with_model: CPU input buffer size = 6.01 MiB
llama_new_context_with_model: CPU compute buffer size = 504.25 MiB
llama_new_context_with_model: graph splits (measure): 1
system_info: n_threads = 24 / 48 | AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 1 | NEON = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 0 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 |
sampling:
repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000
top_k = 40, tfs_z = 1.000, top_p = 0.950, min_p = 0.050, typical_p = 1.000, temp = 0.800
mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000
sampling order:
CFG -> Penalties -> top_k -> tfs_z -> typical_p -> top_p -> min_p -> temperature
generate: n_ctx = 512, n_batch = 512, n_predict = 256, n_keep = 1
Write a Python function to find sum of two numbers.
```python
def sum(a, b):
"""
Finds the sum of two numbers.
Args:
a (int): The first number.
b (int): The second number.
Returns:
int: The sum of a and b.
"""
return a + b
# Get the two numbers from the user.
num1 = int(input("Enter the first number: "))
num2 = int(input("Enter the second number: "))
# Find the sum of num1 and num2.
sum_of_numbers = sum(num1, num2)
# Print the sum of num1 and num2.
print(f"The sum of {num1} and {num2} is {sum_of_numbers}")
``` [end of text]
llama_print_timings: load time = 4991.67 ms
llama_print_timings: sample time = 201.42 ms / 177 runs ( 1.14 ms per token, 878.76 tokens per second)
llama_print_timings: prompt eval time = 222.46 ms / 12 tokens ( 18.54 ms per token, 53.94 tokens per second)
llama_print_timings: eval time = 29505.66 ms / 176 runs ( 167.65 ms per token, 5.96 tokens per second)
llama_print_timings: total time = 30275.88 ms / 188 tokens
Log end
Running the model on a single / multi GPU
llama.cpp/build$ bin/main -m gemma-2b-it.gguf -n 256 -p "Write a Python function to find sum of two numbers." --repeat-penalty 1.1 -ngl 99
Log start
main: build = 2234 (973053d8)
main: built with cc (Debian 13.2.0-5) 13.2.0 for x86_64-linux-gnu
main: seed = 1708975775
ggml_init_cublas: GGML_CUDA_FORCE_MMQ: no
ggml_init_cublas: CUDA_USE_TENSOR_CORES: yes
ggml_init_cublas: found 1 CUDA devices:
Device 0: NVIDIA GeForce RTX 3060, compute capability 8.6, VMM: yes
llama_model_loader: loaded meta data with 19 key-value pairs and 164 tensors from gemma-2b-it.gguf (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv 0: general.architecture str = gemma
llama_model_loader: - kv 1: general.name str = gemma-2b-it
llama_model_loader: - kv 2: gemma.context_length u32 = 8192
llama_model_loader: - kv 3: gemma.block_count u32 = 18
llama_model_loader: - kv 4: gemma.embedding_length u32 = 2048
llama_model_loader: - kv 5: gemma.feed_forward_length u32 = 16384
llama_model_loader: - kv 6: gemma.attention.head_count u32 = 8
llama_model_loader: - kv 7: gemma.attention.head_count_kv u32 = 1
llama_model_loader: - kv 8: gemma.attention.key_length u32 = 256
llama_model_loader: - kv 9: gemma.attention.value_length u32 = 256
llama_model_loader: - kv 10: gemma.attention.layer_norm_rms_epsilon f32 = 0.000001
llama_model_loader: - kv 11: tokenizer.ggml.model str = llama
llama_model_loader: - kv 12: tokenizer.ggml.bos_token_id u32 = 2
llama_model_loader: - kv 13: tokenizer.ggml.eos_token_id u32 = 1
llama_model_loader: - kv 14: tokenizer.ggml.padding_token_id u32 = 0
llama_model_loader: - kv 15: tokenizer.ggml.unknown_token_id u32 = 3
llama_model_loader: - kv 16: tokenizer.ggml.tokens arr[str,256128] = ["<pad>", "<eos>", "<bos>", "<unk>", ...
llama_model_loader: - kv 17: tokenizer.ggml.scores arr[f32,256128] = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv 18: tokenizer.ggml.token_type arr[i32,256128] = [3, 3, 3, 2, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - type f32: 164 tensors
llm_load_vocab: mismatch in special tokens definition ( 544/256128 vs 388/256128 ).
llm_load_print_meta: format = GGUF V3 (latest)
llm_load_print_meta: arch = gemma
llm_load_print_meta: vocab type = SPM
llm_load_print_meta: n_vocab = 256128
llm_load_print_meta: n_merges = 0
llm_load_print_meta: n_ctx_train = 8192
llm_load_print_meta: n_embd = 2048
llm_load_print_meta: n_head = 8
llm_load_print_meta: n_head_kv = 1
llm_load_print_meta: n_layer = 18
llm_load_print_meta: n_rot = 256
llm_load_print_meta: n_embd_head_k = 256
llm_load_print_meta: n_embd_head_v = 256
llm_load_print_meta: n_gqa = 8
llm_load_print_meta: n_embd_k_gqa = 256
llm_load_print_meta: n_embd_v_gqa = 256
llm_load_print_meta: f_norm_eps = 0.0e+00
llm_load_print_meta: f_norm_rms_eps = 1.0e-06
llm_load_print_meta: f_clamp_kqv = 0.0e+00
llm_load_print_meta: f_max_alibi_bias = 0.0e+00
llm_load_print_meta: n_ff = 16384
llm_load_print_meta: n_expert = 0
llm_load_print_meta: n_expert_used = 0
llm_load_print_meta: rope scaling = linear
llm_load_print_meta: freq_base_train = 10000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_yarn_orig_ctx = 8192
llm_load_print_meta: rope_finetuned = unknown
llm_load_print_meta: model type = 2B
llm_load_print_meta: model ftype = all F32 (guessed)
llm_load_print_meta: model params = 2.51 B
llm_load_print_meta: model size = 9.34 GiB (32.00 BPW)
llm_load_print_meta: general.name = gemma-2b-it
llm_load_print_meta: BOS token = 2 '<bos>'
llm_load_print_meta: EOS token = 1 '<eos>'
llm_load_print_meta: UNK token = 3 '<unk>'
llm_load_print_meta: PAD token = 0 '<pad>'
llm_load_print_meta: LF token = 227 '<0x0A>'
llm_load_tensors: ggml ctx size = 0.13 MiB
llm_load_tensors: offloading 18 repeating layers to GPU
llm_load_tensors: offloading non-repeating layers to GPU
llm_load_tensors: offloaded 19/19 layers to GPU
llm_load_tensors: CPU buffer size = 2001.00 MiB
llm_load_tensors: CUDA0 buffer size = 9561.29 MiB
.............................................................
llama_new_context_with_model: n_ctx = 512
llama_new_context_with_model: freq_base = 10000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init: CUDA0 KV buffer size = 9.00 MiB
llama_new_context_with_model: KV self size = 9.00 MiB, K (f16): 4.50 MiB, V (f16): 4.50 MiB
llama_new_context_with_model: CUDA_Host input buffer size = 6.01 MiB
ggml_gallocr_reserve_n: reallocating CUDA0 buffer from size 0.00 MiB to 508.25 MiB
ggml_gallocr_reserve_n: reallocating CUDA_Host buffer from size 0.00 MiB to 4.00 MiB
llama_new_context_with_model: CUDA0 compute buffer size = 508.25 MiB
llama_new_context_with_model: CUDA_Host compute buffer size = 4.00 MiB
llama_new_context_with_model: graph splits (measure): 3
ggml_gallocr_needs_realloc: graph has different number of nodes
ggml_gallocr_alloc_graph: cannot reallocate multi buffer graph automatically, call reserve
ggml_backend_sched: failed to allocate graph, reserving
system_info: n_threads = 6 / 12 | AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 1 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 1 | NEON = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 |
sampling:
repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000
top_k = 40, tfs_z = 1.000, top_p = 0.950, min_p = 0.050, typical_p = 1.000, temp = 0.800
mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000
sampling order:
CFG -> Penalties -> top_k -> tfs_z -> typical_p -> top_p -> min_p -> temperature
generate: n_ctx = 512, n_batch = 512, n_predict = 256, n_keep = 1
Write a Python function to find sum of two numbers.ggml_gallocr_needs_realloc: node inp_embd is not valid
ggml_gallocr_alloc_graph: cannot reallocate multi buffer graph automatically, call reserve
ggml_backend_sched: failed to allocate graph, reserving
```python
def sum(a, b):
return a + b
# Get the two numbers from the user
num1 = int(input("Enter first number: "))
num2 = int(input("Enter second number: "))
# Calculate and print the sum
sum_of_numbers = sum(num1, num2)
print(f"Sum of {num1} and {num2} is {sum_of_numbers}")
``` [end of text]
llama_print_timings: load time = 12672.17 ms
llama_print_timings: sample time = 2352.84 ms / 101 runs ( 23.30 ms per token, 42.93 tokens per second)
llama_print_timings: prompt eval time = 35.28 ms / 12 tokens ( 2.94 ms per token, 340.16 tokens per second)
llama_print_timings: eval time = 3189.28 ms / 100 runs ( 31.89 ms per token, 31.36 tokens per second)
llama_print_timings: total time = 5898.47 ms / 112 tokens
Log end
Inputs and outputs
- Input: Text string, such as a question, a prompt, or a document to be summarized.
- Output: Generated English-language text in response to the input, such as an answer to a question, or a summary of a document.
Model Data
Data used for model training and how the data was processed.
Training Dataset
These models were trained on a dataset of text data that includes a wide variety of sources, totaling 6 trillion tokens. Here are the key components:
- Web Documents: A diverse collection of web text ensures the model is exposed to a broad range of linguistic styles, topics, and vocabulary. Primarily English-language content.
- Code: Exposing the model to code helps it to learn the syntax and patterns of programming languages, which improves its ability to generate code or understand code-related questions.
- Mathematics: Training on mathematical text helps the model learn logical reasoning, symbolic representation, and to address mathematical queries.
The combination of these diverse data sources is crucial for training a powerful language model that can handle a wide variety of different tasks and text formats.
Data Preprocessing
Here are the key data cleaning and filtering methods applied to the training data:
- CSAM Filtering: Rigorous CSAM (Child Sexual Abuse Material) filtering was applied at multiple stages in the data preparation process to ensure the exclusion of harmful and illegal content
- Sensitive Data Filtering: As part of making Gemma pre-trained models safe and reliable, automated techniques were used to filter out certain personal information and other sensitive data from training sets.
- Additional methods: Filtering based on content quality and safely in line with our policies.
Implementation Information
Details about the model internals.
Hardware
Gemma was trained using the latest generation of Tensor Processing Unit (TPU) hardware (TPUv5e).
Training large language models requires significant computational power. TPUs, designed specifically for matrix operations common in machine learning, offer several advantages in this domain:
- Performance: TPUs are specifically designed to handle the massive computations involved in training LLMs. They can speed up training considerably compared to CPUs.
- Memory: TPUs often come with large amounts of high-bandwidth memory, allowing for the handling of large models and batch sizes during training. This can lead to better model quality.
- Scalability: TPU Pods (large clusters of TPUs) provide a scalable solution for handling the growing complexity of large foundation models. You can distribute training across multiple TPU devices for faster and more efficient processing.
- Cost-effectiveness: In many scenarios, TPUs can provide a more cost-effective solution for training large models compared to CPU-based infrastructure, especially when considering the time and resources saved due to faster training.
- These advantages are aligned with Google's commitments to operate sustainably.
Software
Training was done using JAX and ML Pathways.
JAX allows researchers to take advantage of the latest generation of hardware, including TPUs, for faster and more efficient training of large models.
ML Pathways is Google's latest effort to build artificially intelligent systems capable of generalizing across multiple tasks. This is specially suitable for foundation models, including large language models like these ones.
Together, JAX and ML Pathways are used as described in the paper about the Gemini family of models; "the 'single controller' programming model of Jax and Pathways allows a single Python process to orchestrate the entire training run, dramatically simplifying the development workflow."
Evaluation
Model evaluation metrics and results.
Benchmark Results
These models were evaluated against a large collection of different datasets and metrics to cover different aspects of text generation:
| Benchmark | Metric | 2B Params | 7B Params |
|---|---|---|---|
| MMLU | 5-shot, top-1 | 42.3 | 64.3 |
| HellaSwag | 0-shot | 71.4 | 81.2 |
| PIQA | 0-shot | 77.3 | 81.2 |
| SocialIQA | 0-shot | 59.7 | 51.8 |
| BooIQ | 0-shot | 69.4 | 83.2 |
| WinoGrande | partial score | 65.4 | 72.3 |
| CommonsenseQA | 7-shot | 65.3 | 71.3 |
| OpenBookQA | 47.8 | 52.8 | |
| ARC-e | 73.2 | 81.5 | |
| ARC-c | 42.1 | 53.2 | |
| TriviaQA | 5-shot | 53.2 | 63.4 |
| Natural Questions | 5-shot | - | 23 |
| HumanEval | pass@1 | 22.0 | 32.3 |
| MBPP | 3-shot | 29.2 | 44.4 |
| GSM8K | maj@1 | 17.7 | 46.4 |
| MATH | 4-shot | 11.8 | 24.3 |
| AGIEval | 24.2 | 41.7 | |
| BIG-Bench | 35.2 | 55.1 | |
| ------------------------------ | ------------- | ----------- | --------- |
| Average | 54.0 | 56.4 |
Ethics and Safety
Ethics and safety evaluation approach and results.
Evaluation Approach
Our evaluation methods include structured evaluations and internal red-teaming testing of relevant content policies. Red-teaming was conducted by a number of different teams, each with different goals and human evaluation metrics. These models were evaluated against a number of different categories relevant to ethics and safety, including:
- Text-to-Text Content Safety: Human evaluation on prompts covering safety policies including child sexual abuse and exploitation, harassment, violence and gore, and hate speech.
- Text-to-Text Representational Harms: Benchmark against relevant academic datasets such as WinoBias and BBQ Dataset.
- Memorization: Automated evaluation of memorization of training data, including the risk of personally identifiable information exposure.
- Large-scale harm: Tests for "dangerous capabilities," such as chemical, biological, radiological, and nuclear (CBRN) risks.
Evaluation Results
The results of ethics and safety evaluations are within acceptable thresholds for meeting internal policies for categories such as child safety, content safety, representational harms, memorization, large-scale harms. On top of robust internal evaluations, the results of well known safety benchmarks like BBQ, BOLD, Winogender, Winobias, RealToxicity, and TruthfulQA are shown here.
| Benchmark | Metric | 2B Params | 7B Params |
|---|---|---|---|
| RealToxicity | average | 6.86 | 7.90 |
| BOLD | 45.57 | 49.08 | |
| CrowS-Pairs | top-1 | 45.82 | 51.33 |
| BBQ Ambig | 1-shot, top-1 | 62.58 | 92.54 |
| BBQ Disambig | top-1 | 54.62 | 71.99 |
| Winogender | top-1 | 51.25 | 54.17 |
| TruthfulQA | 44.84 | 31.81 | |
| Winobias 1_2 | 56.12 | 59.09 | |
| Winobias 2_2 | 91.10 | 92.23 | |
| Toxigen | 29.77 | 39.59 | |
| ------------------------------ | ------------- | ----------- | --------- |
Usage and Limitations
These models have certain limitations that users should be aware of.
Intended Usage
Open Large Language Models (LLMs) have a wide range of applications across various industries and domains. The following list of potential uses is not comprehensive. The purpose of this list is to provide contextual information about the possible use-cases that the model creators considered as part of model training and development.
- Content Creation and Communication
- Text Generation: These models can be used to generate creative text formats such as poems, scripts, code, marketing copy, and email drafts.
- Chatbots and Conversational AI: Power conversational interfaces for customer service, virtual assistants, or interactive applications.
- Text Summarization: Generate concise summaries of a text corpus, research papers, or reports.
- Research and Education
- Natural Language Processing (NLP) Research: These models can serve as a foundation for researchers to experiment with NLP techniques, develop algorithms, and contribute to the advancement of the field.
- Language Learning Tools: Support interactive language learning experiences, aiding in grammar correction or providing writing practice.
- Knowledge Exploration: Assist researchers in exploring large bodies of text by generating summaries or answering questions about specific topics.
Limitations
- Training Data
- The quality and diversity of the training data significantly influence the model's capabilities. Biases or gaps in the training data can lead to limitations in the model's responses.
- The scope of the training dataset determines the subject areas the model can handle effectively.
- Context and Task Complexity
- LLMs are better at tasks that can be framed with clear prompts and instructions. Open-ended or highly complex tasks might be challenging.
- A model's performance can be influenced by the amount of context provided (longer context generally leads to better outputs, up to a certain point).
- Language Ambiguity and Nuance
- Natural language is inherently complex. LLMs might struggle to grasp subtle nuances, sarcasm, or figurative language.
- Factual Accuracy
- LLMs generate responses based on information they learned from their training datasets, but they are not knowledge bases. They may generate incorrect or outdated factual statements.
- Common Sense
- LLMs rely on statistical patterns in language. They might lack the ability to apply common sense reasoning in certain situations.
Ethical Considerations and Risks
The development of large language models (LLMs) raises several ethical concerns. In creating an open model, we have carefully considered the following:
- Bias and Fairness
- LLMs trained on large-scale, real-world text data can reflect socio-cultural biases embedded in the training material. These models underwent careful scrutiny, input data pre-processing described and posterior evaluations reported in this card.
- Misinformation and Misuse
- LLMs can be misused to generate text that is false, misleading, or harmful.
- Guidelines are provided for responsible use with the model, see the Responsible Generative AI Toolkit.
- Transparency and Accountability:
- This model card summarizes details on the models' architecture, capabilities, limitations, and evaluation processes.
- A responsibly developed open model offers the opportunity to share innovation by making LLM technology accessible to developers and researchers across the AI ecosystem.
Risks identified and mitigations:
- Perpetuation of biases: It's encouraged to perform continuous monitoring (using evaluation metrics, human review) and the exploration of de-biasing techniques during model training, fine-tuning, and other use cases.
- Generation of harmful content: Mechanisms and guidelines for content safety are essential. Developers are encouraged to exercise caution and implement appropriate content safety safeguards based on their specific product policies and application use cases.
- Misuse for malicious purposes: Technical limitations and developer and end-user education can help mitigate against malicious applications of LLMs. Educational resources and reporting mechanisms for users to flag misuse are provided. Prohibited uses of Gemma models are outlined in the Gemma Prohibited Use Policy.
- Privacy violations: Models were trained on data filtered for removal of PII (Personally Identifiable Information). Developers are encouraged to adhere to privacy regulations with privacy-preserving techniques.
Benefits
At the time of release, this family of models provides high-performance open large language model implementations designed from the ground up for Responsible AI development compared to similarly sized models.
Using the benchmark evaluation metrics described in this document, these models have shown to provide superior performance to other, comparably-sized open model alternatives.