Skip to content

GitLab

"tests/test_codebase/test_mmseg/__init__.py" did not exist on "a17c53b87176187619e207516973054a28c59335"
Commit 5d3bf2e7 authored by lintangsutawika's avatar lintangsutawika
Browse files

Merge branch 'big-refactor' of... 

Merge branch 'big-refactor' of https://github.com/EleutherAI/lm-evaluation-harness into openai_completions
parents f66730c4 bf26d979
Hide whitespace changes
Inline Side-by-side
Showing with 637 additions and 121 deletions
README.md
View file @ 5d3bf2e7
...@@ -4,17 +4,16 @@ ...@@ -4,17 +4,16 @@
This project provides a unified framework to test generative language models on a large number of different evaluation tasks. This project provides a unified framework to test generative language models on a large number of different evaluation tasks.
Features: **Features:**
- Over 60 standard academic benchmarks for LLMs, with hundreds of subtasks and variants implemented. - Over 60 standard academic benchmarks for LLMs, with hundreds of subtasks and variants implemented.
- Support for models loaded via [transformers](https://github.com/huggingface/transformers/) (including quantization via [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ)), [GPT-NeoX](https://github.com/EleutherAI/gpt-neox), and [Megatron-DeepSpeed](https://github.com/microsoft/Megatron-DeepSpeed/), with a flexible tokenization-agnostic interface. - Support for models loaded via [transformers](https://github.com/huggingface/transformers/) (including quantization via [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ)), [GPT-NeoX](https://github.com/EleutherAI/gpt-neox), and [Megatron-DeepSpeed](https://github.com/microsoft/Megatron-DeepSpeed/), with a flexible tokenization-agnostic interface.
- Support for commercial APIs including [OpenAI](https://openai.com), [goose.ai](https://goose.ai), and [TextSynth](https://textsynth.com/). - Support for commercial APIs including [OpenAI](https://openai.com), [goose.ai](https://goose.ai), and [TextSynth](https://textsynth.com/).
- Support for evaluation on adapters (e.g. LoRA) supported in [HuggingFace's PEFT library](https://github.com/huggingface/peft). - Support for evaluation on adapters (e.g. LoRA) supported in [HuggingFace's PEFT library](https://github.com/huggingface/peft).
- Support for local models and benchmarks. - Support for local models and benchmarks.
- Evaluation with publicly available prompts ensures reproducibility and comparability between papers. - Evaluation with publicly available prompts ensures reproducibility and comparability between papers.
- Easy support for custom prompts and evaluation metrics.
The Language Model Evaluation Harness is the backend for 🤗 Hugging Face's popular [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard) and is used internally by dozens of companies including NVIDIA, Cohere, Booz Allen Hamilton, and Mosaic ML. The Language Model Evaluation Harness is the backend for 🤗 Hugging Face's popular [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard), has been used in [hundreds of papers](https://scholar.google.com/scholar?oi=bibs&hl=en&authuser=2&cites=15052937328817631261,4097184744846514103,17476825572045927382,18443729326628441434,12854182577605049984) is used internally by dozens of companies including NVIDIA, Cohere, Booz Allen Hamilton, and Mosaic ML.
## Install ## Install
...@@ -26,25 +25,22 @@ cd lm-evaluation-harness ...@@ -26,25 +25,22 @@ cd lm-evaluation-harness
pip install -e . pip install -e .
``` ```
To install additional multilingual tokenization and text segmentation packages, you must install the package with the `multilingual` extra: We also provide a number of optional dependencies for . Extras can be installed via `pip install -e ".[NAME]"`
```bash
pip install -e ".[multilingual]"
```
To support loading GPTQ quantized models, install the package with the `gptq` extra:
```bash | Name | Use |
pip install -e ".[gptq]" | ------------- | ------------------------------------- |
``` | anthropic | For using Anthropic's models |
| dev | You probably don't want to use this |
| gptq | For loading models with GPTQ |
| testing | You probably don't want to use this |
| multilingual | For multilingual tokenizers |
| openai | For using OpenAI's models |
| promptsource | For using PromtSource prompts |
| sentencepiece | For using the sentencepiece tokenizer |
| vllm | For loading models with vLLM |
| all | Loads all extras |
### Support
To install the package with all extras, run
```bash
pip install -e ".[all]"
```
## Support
The best way to get support is to open an issue on this repo or join the EleutherAI discord server](discord.gg/eleutherai). The `#lm-thunderdome` channel is dedicated to developing this project and the `#release-discussion` channel is for receiving support for our releases. The best way to get support is to open an issue on this repo or join the EleutherAI discord server](discord.gg/eleutherai). The `#lm-thunderdome` channel is dedicated to developing this project and the `#release-discussion` channel is for receiving support for our releases.
...@@ -54,7 +50,6 @@ The best way to get support is to open an issue on this repo or join the Eleuthe ...@@ -54,7 +50,6 @@ The best way to get support is to open an issue on this repo or join the Eleuthe
To evaluate a model hosted on the [HuggingFace Hub](https://huggingface.co/models) (e.g. GPT-J-6B) on `hellaswag` you can use the following command: To evaluate a model hosted on the [HuggingFace Hub](https://huggingface.co/models) (e.g. GPT-J-6B) on `hellaswag` you can use the following command:
```bash ```bash
python -m lm_eval \ python -m lm_eval \
--model hf \ --model hf \
...@@ -92,9 +87,7 @@ Alternatively, you can use `lm-eval` or `lm_eval` instead of `python -m lm_eval` ...@@ -92,9 +87,7 @@ Alternatively, you can use `lm-eval` or `lm_eval` instead of `python -m lm_eval`
### Multi-GPU Evaluation with Hugging Face `accelerate` ### Multi-GPU Evaluation with Hugging Face `accelerate`
To parallelize evaluation of HuggingFace models across multiple GPUs, we allow for two different types of multi-GPU evaluation. To parallelize evaluation of HuggingFace models across multiple GPUs, we leverage the [accelerate 🚀](https://github.com/huggingface/accelerate) library as follows:
The first is performed by launching evaluation via the `accelerate` library as follows:
``` ```
accelerate launch -m lm_eval \ accelerate launch -m lm_eval \
...@@ -107,48 +100,55 @@ This will perform *data-parallel evaluation*: that is, placing a **single full c ...@@ -107,48 +100,55 @@ This will perform *data-parallel evaluation*: that is, placing a **single full c
If your model is *is too large to be run on a single one of your GPUs* then you can use `accelerate` with Fully Sharded Data Parallel (FSDP) that splits the weights of the model across your data parallel ranks. To enable this, ensure you select `YES` when asked ```Do you want to use FullyShardedDataParallel?``` when running `accelerate config`. To enable memory-efficient loading, select `YES` when asked `Do you want each individually wrapped FSDP unit to broadcast module parameters from rank 0 at the start?`. This will ensure only the rank 0 process loads the model and then broadcasts the parameters to the other ranks instead of having each rank load all parameters which can lead to large RAM usage spikes around the start of the script that may cause errors. If your model is *is too large to be run on a single one of your GPUs* then you can use `accelerate` with Fully Sharded Data Parallel (FSDP) that splits the weights of the model across your data parallel ranks. To enable this, ensure you select `YES` when asked ```Do you want to use FullyShardedDataParallel?``` when running `accelerate config`. To enable memory-efficient loading, select `YES` when asked `Do you want each individually wrapped FSDP unit to broadcast module parameters from rank 0 at the start?`. This will ensure only the rank 0 process loads the model and then broadcasts the parameters to the other ranks instead of having each rank load all parameters which can lead to large RAM usage spikes around the start of the script that may cause errors.
We also provide an second method to run these large models: use of the `parallelize` argument.
```
python -m lm_eval \
--model hf \
--model_args pretrained=EleutherAI/pythia-12b,parallelize=True
--tasks lambada_openai,arc_easy \
--batch_size 16
```
To pass even more advanced keyword arguments to `accelerate`, we allow for the following arguments as well: To pass even more advanced keyword arguments to `accelerate`, we allow for the following arguments as well:
- `device_map_option`: How to split model weights across available GPUs. defaults to "auto". - `device_map_option`: How to split model weights across available GPUs. defaults to "auto".
- `max_memory_per_gpu`: the max GPU memory to use per GPU in loading the model. - `max_memory_per_gpu`: the max GPU memory to use per GPU in loading the model.
- `max_cpu_memory`: the max amount of CPU memory to use when offloading the model weights to RAM. - `max_cpu_memory`: the max amount of CPU memory to use when offloading the model weights to RAM.
- `offload_folder`: a folder where model weights will be offloaded to disk if needed. - `offload_folder`: a folder where model weights will be offloaded to disk if needed.
Note that this method naively splits models across GPUs, resulting in only a single GPU performing work at any point in time, and so is much slower than launching with `accelerate launch`, possibly by a factor of the total # of GPUs.
**Note that this option requires launching evaluation via `python -m lm_eval` rather than `accelerate launch -m lm_eval`.**
To use `accelerate` with the `lm-eval` command, use To use `accelerate` with the `lm-eval` command, use
``` ```
accelerate launch --no_python lm-eval --model ... accelerate launch --no_python lm-eval --model ...
``` ```
### Commercial APIs #### Tensor Parallel + Optimized Inference with vLLM
We also support vLLM for faster inference on [supported model types](https://docs.vllm.ai/en/latest/models/supported_models.html).
To run with vLLM, first install the vllm library, externally or via the lm_eval[vllm] extra:
```bash
pip install -e .[vllm]
```
Then, you can run the library as normal, for single-GPU or tensor-parallel inference, for example:
```bash
python -m lm_eval \
--model vllm \
--model_args pretrained={model_name},tensor_parallel_size={number of GPUs to use},dtype=auto,gpu_memory_utilization=0.8
--tasks lambada_openai
--batch_size auto
```
For a full list of supported vLLM configurations, please reference our vLLM integration and the vLLM documentation.
Our library also supports the evaluation of models served via several commercial APIs, and hope to implement support for common performant local/self-hosted inference servers. ### Supported APIs and Inference Libraries
Our library also supports the evaluation of models served via several commercial APIs, and we hope to implement support for the most commonly used performant local/self-hosted inference servers.
A full accounting of the supported and planned libraries + APIs can be seen below: A full accounting of the supported and planned libraries + APIs can be seen below:
| API or Inference Server | Implemented? | `--model <xxx>` name | Models supported: | Request Types: | | API or Inference Server | Implemented? | `--model <xxx>` name | Models supported: | Request Types: |
|-----------------------------|---------------------------------|----------------------------------------------------------------------------------|--------------------------------------|----------------------------------------------------------| |-----------------------------|---------------------------------|----------------------------------------------------------------------------------|-----------------------------------------------------------------------------------------------|----------------------------------------------------------|
| OpenAI Completions | :heavy_check_mark: | `openai`, `openai-completions`, `gooseai` | up to `code-davinci-002` | `generate_until`, `loglikelihood`, `loglikelihood_rolling` | | OpenAI Completions | :heavy_check_mark: | `openai`, `openai-completions`, `gooseai` | up to `code-davinci-002` | `generate_until`, `loglikelihood`, `loglikelihood_rolling` |
| OpenAI ChatCompletions | :x: Not yet - needs testing! | N/A | [All ChatCompletions API models](https://platform.openai.com/docs/guides/gpt) | `generate_until` (no logprobs) | | OpenAI ChatCompletions | :x: Not yet - needs testing! | N/A | [All ChatCompletions API models](https://platform.openai.com/docs/guides/gpt) | `generate_until` (no logprobs) |
| Anthropic | :heavy_check_mark: | `anthropic` | [Supported Anthropic Engines](https://docs.anthropic.com/claude/reference/selecting-a-model) | `generate_until` (no logprobs) | | Anthropic | :heavy_check_mark: | `anthropic` | [Supported Anthropic Engines](https://docs.anthropic.com/claude/reference/selecting-a-model) | `generate_until` (no logprobs) |
| GooseAI | :heavy_check_mark: (not separately maintained) | `openai`, `openai-completions`, `gooseai` (same interface as OpenAI Completions) | | `generate_until`, `loglikelihood`, `loglikelihood_rolling` | | GooseAI | :heavy_check_mark: (not separately maintained) | `openai`, `openai-completions`, `gooseai` (same interface as OpenAI Completions) | | `generate_until`, `loglikelihood`, `loglikelihood_rolling` |
| Textsynth | Needs testing | `textsynth` | ??? | `generate_until`, `loglikelihood`, `loglikelihood_rolling` | | Textsynth | Needs testing | `textsynth` | ??? | `generate_until`, `loglikelihood`, `loglikelihood_rolling` |
| Cohere | :hourglass: - blocked on Cohere API bug | N/A | [All `cohere.generate()` engines](https://docs.cohere.com/docs/models) | `generate_until`, `loglikelihood`, `loglikelihood_rolling` | | Cohere | [:hourglass: - blocked on Cohere API bug](https://github.com/EleutherAI/lm-evaluation-harness/pull/395) | N/A | [All `cohere.generate()` engines](https://docs.cohere.com/docs/models) | `generate_until`, `loglikelihood`, `loglikelihood_rolling` |
| GGML | :hourglass: [PR](https://github.com/EleutherAI/lm-evaluation-harness/pull/617) | N/A | ??? | `generate_until`, `loglikelihood`, `loglikelihood_rolling` | | GGML/[Llama.cpp](https://github.com/ggerganov/llama.cpp) (via [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)) | :heavy_check_mark: | `gguf`, `ggml` | Llama-architecture models (Llama, Llama 2, Llemma, Mistral(?), Llama finetunes) | `generate_until`, `loglikelihood`, `loglikelihood_rolling` |
| vLLM | :x: Not yet - needs help! | N/A | All HF models | `generate_until` (no logprobs) | | vLLM | :heavy_check_mark: | `vllm` | [Most HF Causal Language Models](https://docs.vllm.ai/en/latest/models/supported_models.html) | `generate_until`, `loglikelihood`, `loglikelihood_rolling` |
| Your inference server here! | ... | ... | ... | ... | | ... | | Your inference server here! | ... | ... | ... | ... | | ... |
It is on our roadmap to create task variants designed to enable models which do not serve logprobs/loglikelihoods to be compared with generation performance of open-source models. It is on our roadmap to create task variants designed to enable models which do not serve logprobs/loglikelihoods to be compared with generation performance of open-source models.
...@@ -194,7 +194,7 @@ python -m lm_eval \ ...@@ -194,7 +194,7 @@ python -m lm_eval \
--check_integrity --check_integrity
``` ```
## Advanced Usage ## Advanced Usage Tips
For models loaded with the HuggingFace `transformers` library, any arguments provided via `--model_args` get passed to the relevant constructor directly. This means that anything you can do with `AutoModel` can be done with our library. For example, you can pass a local path via `pretrained=` or use models finetuned with [PEFT](https://github.com/huggingface/peft) by taking the call you would run to evaluate the base model and add `,peft=PATH` to the `model_args` argument: For models loaded with the HuggingFace `transformers` library, any arguments provided via `--model_args` get passed to the relevant constructor directly. This means that anything you can do with `AutoModel` can be done with our library. For example, you can pass a local path via `pretrained=` or use models finetuned with [PEFT](https://github.com/huggingface/peft) by taking the call you would run to evaluate the base model and add `,peft=PATH` to the `model_args` argument:
```bash ```bash
...@@ -216,21 +216,31 @@ python -m lm_eval \ ...@@ -216,21 +216,31 @@ python -m lm_eval \
We support wildcards in task names, for example you can run all of the machine-translated lambada tasks via `--task lambada_openai_mt_*`. We support wildcards in task names, for example you can run all of the machine-translated lambada tasks via `--task lambada_openai_mt_*`.
To save evaluation results provide an `--output_path`. We also support logging model responses with the `--log_samples` flag for post-hoc analysis.
Additionally, one can provide a directory with `--use_cache` to cache the results of prior runs. This allows you to avoid repeated execution of the same (model, task) pairs for re-scoring.
For a full list of supported arguments, check out the [interface](https://github.com/EleutherAI/lm-evaluation-harness/blob/big-refactor/docs/interface.md) guide in our documentation!
## How to Contribute or Learn More? ## How to Contribute or Learn More?
For more information on the library and how everything fits together, check out all of our [documentation pages](https://github.com/EleutherAI/lm-evaluation-harness/tree/big-refactor/docs)! We plan to post a larger roadmap of desired + planned library improvements soon, with more information on how contributors can help. For more information on the library and how everything fits together, check out all of our [documentation pages](https://github.com/EleutherAI/lm-evaluation-harness/tree/big-refactor/docs)! We plan to post a larger roadmap of desired + planned library improvements soon, with more information on how contributors can help.
You can also ask for help, or discuss new features with the maintainers in the #lm-thunderdome channel of the EleutherAI discord! If you've used the library and have had a positive (or negative) experience, we'd love to hear from you! You can also ask for help, or discuss new features with the maintainers in the #lm-thunderdome channel of the EleutherAI discord! If you've used the library and have had a positive (or negative) experience, we'd love to hear from you!
### Implementing new tasks ### Implementing new tasks
To implement a new task in the eval harness, see [this guide](./docs/new_task_guide.md). To implement a new task in the eval harness, see [this guide](./docs/new_task_guide.md).
In general, we following the following priority list for addressing concerns about prompting and other eval details:
1. If there is widespread agreement among people who train LLMs, use the agreed upon procedure.
2. If there is a clear and unambiguous official implementation, use that procedure.
3. If there is widespread agreement among people who evaluate LLMs, use the agreed upon procedure.
4. If there are multiple common implementations but not universal or widespread agreement, use our preferred option among the common implementations. As before, prioritize choosing from among the implementations found in LLM training papers.
As a start, we currently only support one prompt per task, which we strive to make the "standard" as defined by the benchmark's authors. If you would like to study how varying prompts causes changes in the evaluation score, we support prompts authored in the [Promptsource Library](https://github.com/bigscience-workshop/promptsource/tree/main) as described further in [the task guide](https://github.com/EleutherAI/lm-evaluation-harness/blob/big-refactor/lm_eval/docs/new_task_guide.md) and [the advanced task guide](https://github.com/EleutherAI/lm-evaluation-harness/blob/big-refactor/lm_eval/docs/advanced_task_guide.md) and welcome contributions of novel task templates and task variants. These are guidelines and not rules, and can be overruled in special circumstances.
We try to prioritize agreement with the procedures used by other groups to decrease the harm when people inevitably compare runs across different papers despite our discouragement of the practice. Historically, we also prioritized the implementation from "Language Models are Few Shot Learners" as our original goal was specifically to compare results with that paper.
## Cite as ## Cite as
......
docs/interface.md
View file @ 5d3bf2e7
...@@ -18,6 +18,8 @@ This mode supports a number of command-line arguments, the details of which can ...@@ -18,6 +18,8 @@ This mode supports a number of command-line arguments, the details of which can
* `--num_fewshot` : Sets the number of few-shot examples to place in context. Must be an integer. * `--num_fewshot` : Sets the number of few-shot examples to place in context. Must be an integer.
* `--gen_kwargs` : takes an arg string in same format as `--model_args` and creates a dictionary of keyword arguments. These will be passed to the models for all called `generate_until` (free-form or greedy generation task) tasks, to set options such as the sampling temperature or `top_p` / `top_k`. For a list of what args are supported for each model type, reference the respective library's documentation (for example, the documentation for `transformers.AutoModelForCausalLM.generate()`.) These kwargs will be applied to all `generate_until` tasks called--we do not currently support unique gen_kwargs or batch_size values per task in a single run of the library. To control these on a per-task level, set them in that task's YAML file.
* `--batch_size` : Sets the batch size used for evaluation. Can be a positive integer or `"auto"` to automatically select the largest batch size that will fit in memory, speeding up evaluation. One can pass `--batch_size auto:N` to re-select the maximum batch size `N` times during evaluation. This can help accelerate evaluation further, since `lm-eval` sorts documents in descending order of context length. * `--batch_size` : Sets the batch size used for evaluation. Can be a positive integer or `"auto"` to automatically select the largest batch size that will fit in memory, speeding up evaluation. One can pass `--batch_size auto:N` to re-select the maximum batch size `N` times during evaluation. This can help accelerate evaluation further, since `lm-eval` sorts documents in descending order of context length.
* `--max_batch_size` : Sets the maximum batch size to try to fit in memory, if `--batch_size auto` is passed. * `--max_batch_size` : Sets the maximum batch size to try to fit in memory, if `--batch_size auto` is passed.
......
docs/model_guide.md
View file @ 5d3bf2e7
# New Model Guide # New Model Guide
The `lm-evaluation-harness` is intended to be a model-agnostic framework for evaluating . We provide first-class support for HuggingFace `AutoModelForCausalLM` and `AutoModelForSeq2SeqLM` type models, but
This guide may be of special interest to users who are using the library outside of the repository, via installing the library via pypi and calling `lm_eval.evaluator.evaluate()` to evaluate an existing model. This guide may be of special interest to users who are using the library outside of the repository, via installing the library via pypi and calling `lm_eval.evaluator.evaluate()` to evaluate an existing model.
In order to properly evaluate a given LM, we require implementation of a wrapper class subclassing the `lm_eval.api.model.LM` class, that defines how the Evaluation Harness should interface with your model. This guide walks through how to write this `LM` subclass via adding it to the library! In order to properly evaluate a given LM, we require implementation of a wrapper class subclassing the `lm_eval.api.model.LM` class, that defines how the Evaluation Harness should interface with your model. This guide walks through how to write this `LM` subclass via adding it to the library!
...@@ -104,6 +102,8 @@ class MyCustomLM(LM): ...@@ -104,6 +102,8 @@ class MyCustomLM(LM):
Using this decorator results in the class being added to an accounting of the usable LM types maintained internally to the library at `lm_eval.api.registry.MODEL_REGISTRY`. See `lm_eval.api.registry` for more detail on what sorts of registries and decorators exist in the library! Using this decorator results in the class being added to an accounting of the usable LM types maintained internally to the library at `lm_eval.api.registry.MODEL_REGISTRY`. See `lm_eval.api.registry` for more detail on what sorts of registries and decorators exist in the library!
**Tip: be sure to import your model in `lm_eval/models/__init__.py!`**
## Testing ## Testing
We also recommend that new model contributions be accompanied by short tests of their 3 core functionalities, at minimum. To see an example of such tests, look at https://github.com/EleutherAI/lm-evaluation-harness/blob/35bdecd379c0cefad6897e67db892f4a6026a128/tests/test_ggml.py . We also recommend that new model contributions be accompanied by short tests of their 3 core functionalities, at minimum. To see an example of such tests, look at https://github.com/EleutherAI/lm-evaluation-harness/blob/35bdecd379c0cefad6897e67db892f4a6026a128/tests/test_ggml.py .
......
examples/README.md deleted 100644 → 0
View file @ f66730c4
This folder is meant to contain instructions and task setups required to evaluate certain papers which may perform non-standard evaluation setups.
Tasks can be supported already in the library under `lm_eval/tasks`, or if highly paper-specific, may remain as YAMLs in the respective `examples/paper-title` folder.
## Verified Papers:
* [WIP] [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://arxiv.org/abs/2201.11903)
* Further details can be found in the `chain_of_thought` subfolder.
## Candidates to Support:
* Least-to-Most Prompting
* Algorithmic Prompting
* Other in-scope prompting techniques
* Multi-turn prompting strategies are likely out of scope for the repository.
* Pythia Suite: Term Frequencies over training
* All setups from GPT-3 Paper
* Varying few-shot orderings + selection ; Varying the label choices for multiple-choice tasks
* Your Paper Here!
examples/chain_of_thought/README.md deleted 100644 → 0
View file @ f66730c4
# Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
https://arxiv.org/abs/2201.11903
## All Tasks in Paper
* ...
* ...
* ...
## Reproduction Scripts
* ...
lm_eval/__main__.py
View file @ 5d3bf2e7
...@@ -105,6 +105,14 @@ def parse_eval_args() -> argparse.Namespace: ...@@ -105,6 +105,14 @@ def parse_eval_args() -> argparse.Namespace:
default=None, default=None,
help="Additional path to include if there are external tasks to include.", help="Additional path to include if there are external tasks to include.",
) )
parser.add_argument(
"--gen_kwargs",
default="",
help=(
"String arguments for model generation on greedy_until tasks,"
" e.g. `temperature=0,top_k=0,top_p=0`"
),
)
parser.add_argument( parser.add_argument(
"--verbosity", "--verbosity",
type=str, type=str,
...@@ -210,6 +218,7 @@ def cli_evaluate(args: Union[argparse.Namespace, None] = None) -> None: ...@@ -210,6 +218,7 @@ def cli_evaluate(args: Union[argparse.Namespace, None] = None) -> None:
check_integrity=args.check_integrity, check_integrity=args.check_integrity,
write_out=args.write_out, write_out=args.write_out,
log_samples=args.log_samples, log_samples=args.log_samples,
gen_kwargs=args.gen_kwargs,
) )
if results is not None: if results is not None:
...@@ -236,7 +245,7 @@ def cli_evaluate(args: Union[argparse.Namespace, None] = None) -> None: ...@@ -236,7 +245,7 @@ def cli_evaluate(args: Union[argparse.Namespace, None] = None) -> None:
filename.open("w").write(samples_dumped) filename.open("w").write(samples_dumped)
print( print(
f"{args.model} ({args.model_args}), limit: {args.limit}, num_fewshot: {args.num_fewshot}, " f"{args.model} ({args.model_args}), gen_kwargs: ({args.gen_kwargs}), limit: {args.limit}, num_fewshot: {args.num_fewshot}, "
f"batch_size: {args.batch_size}{f' ({batch_sizes})' if batch_sizes else ''}" f"batch_size: {args.batch_size}{f' ({batch_sizes})' if batch_sizes else ''}"
) )
print(evaluator.make_table(results)) print(evaluator.make_table(results))
......
lm_eval/evaluator.py
View file @ 5d3bf2e7
...@@ -20,6 +20,7 @@ from lm_eval.utils import ( ...@@ -20,6 +20,7 @@ from lm_eval.utils import (
make_table, make_table,
create_iterator, create_iterator,
get_git_commit_hash, get_git_commit_hash,
simple_parse_args_string,
eval_logger, eval_logger,
) )
...@@ -40,6 +41,7 @@ def simple_evaluate( ...@@ -40,6 +41,7 @@ def simple_evaluate(
decontamination_ngrams_path=None, decontamination_ngrams_path=None,
write_out: bool = False, write_out: bool = False,
log_samples: bool = True, log_samples: bool = True,
gen_kwargs: str = None,
): ):
"""Instantiate and evaluate a model on a list of tasks. """Instantiate and evaluate a model on a list of tasks.
...@@ -70,6 +72,9 @@ def simple_evaluate( ...@@ -70,6 +72,9 @@ def simple_evaluate(
If True, write out an example document and model input for checking task integrity If True, write out an example document and model input for checking task integrity
:param log_samples: bool :param log_samples: bool
If True, write out all model outputs and documents for per-sample measurement and post-hoc analysis If True, write out all model outputs and documents for per-sample measurement and post-hoc analysis
:param gen_kwargs: str
String arguments for model generation
Ignored for all tasks with loglikelihood output_type
:return :return
Dictionary of results Dictionary of results
""" """
...@@ -83,6 +88,14 @@ def simple_evaluate( ...@@ -83,6 +88,14 @@ def simple_evaluate(
tasks != [] tasks != []
), "No tasks specified, or no tasks found. Please verify the task names." ), "No tasks specified, or no tasks found. Please verify the task names."
if gen_kwargs is not None:
gen_kwargs = simple_parse_args_string(gen_kwargs)
eval_logger.warning(
f"generation_kwargs specified through cli, these settings will be used over set parameters in yaml tasks."
)
if gen_kwargs == "":
gen_kwargs = None
if isinstance(model, str): if isinstance(model, str):
if model_args is None: if model_args is None:
model_args = "" model_args = ""
...@@ -117,6 +130,9 @@ def simple_evaluate( ...@@ -117,6 +130,9 @@ def simple_evaluate(
continue continue
config = task_obj._config config = task_obj._config
if config["output_type"] == "generate_until" and gen_kwargs is not None:
config["generation_kwargs"].update(gen_kwargs)
if num_fewshot is not None: if num_fewshot is not None:
if config["num_fewshot"] > 0: if config["num_fewshot"] > 0:
default_num_fewshot = config["num_fewshot"] default_num_fewshot = config["num_fewshot"]
...@@ -154,6 +170,7 @@ def simple_evaluate( ...@@ -154,6 +170,7 @@ def simple_evaluate(
"use_cache": use_cache, "use_cache": use_cache,
"limit": limit, "limit": limit,
"bootstrap_iters": bootstrap_iters, "bootstrap_iters": bootstrap_iters,
"gen_kwargs": gen_kwargs,
} }
results["git_hash"] = get_git_commit_hash() results["git_hash"] = get_git_commit_hash()
return results return results
......
lm_eval/models/__init__.py
View file @ 5d3bf2e7
...@@ -3,6 +3,7 @@ from . import openai_completions ...@@ -3,6 +3,7 @@ from . import openai_completions
from . import textsynth from . import textsynth
from . import dummy from . import dummy
from . import anthropic_llms from . import anthropic_llms
from . import gguf
from . import vllm_causallms
# TODO: implement __all__ # TODO: implement __all__
lm_eval/models/gguf.py 0 → 100644
View file @ 5d3bf2e7
import requests
import logging
import time
from tqdm import tqdm
from requests.exceptions import RequestException
from lm_eval.api.model import LM
from lm_eval.api.registry import register_model
logger = logging.getLogger(__name__)
def get_result(logprobs, context_length):
is_greedy = True
offsets = logprobs["text_offset"]
tokens = logprobs["tokens"]
tokens_logprobs = logprobs["token_logprobs"]
idx = 0
while offsets[idx] < context_length:
idx += 1
continuation_logprobs = sum(tokens_logprobs[idx:-1])
for i in range(idx, len(tokens)):
token = tokens[i]
top_tokens = logprobs["top_logprobs"][i]
top_token = max(top_tokens.keys(), key=lambda x: top_tokens[x])
if top_token != token:
is_greedy = False
break
return continuation_logprobs, is_greedy
@register_model("gguf", "ggml")
class GGUFLM(LM):
def __init__(self, base_url=None, max_length=2048, **kwargs):
super().__init__()
self.base_url = base_url
assert self.base_url, "must pass `base_url` to use GGUF LM!"
self.logprobs = 10
self.temperature = 0.0
self.max_length = max_length
def gguf_completion(
self, context, continuation=None, stop=None, retries=3, delay=5, **kwargs
):
for _ in range(retries):
try:
prompt = context
request = {
"prompt": prompt,
"logprobs": self.logprobs,
"temperature": self.temperature,
}
if continuation:
prompt += continuation
request.update({"prompt": prompt, "max_tokens": 1, "echo": True})
if stop is not None:
request["stop"] = stop
response = requests.post(
f"{self.base_url}/v1/completions", json=request
)
response.raise_for_status()
return response.json()
except RequestException as e:
logger.error(f"RequestException: {e}")
time.sleep(delay) # wait before retrying
else:
raise Exception(f"Failed to get a valid response after {retries} retries.")
def loglikelihood(self, requests):
if not requests:
return []
res = []
for context, continuation in tqdm([req.args for req in requests]):
response = self.gguf_completion(context=context, continuation=continuation)
if response and "choices" in response and response["choices"]:
choice = response["choices"][0]
logprobs = choice.get("logprobs")
if (
logprobs
and "token_logprobs" in logprobs
and logprobs["token_logprobs"]
):
logprob, is_greedy = get_result(logprobs, len(context))
res.append((logprob, is_greedy))
else:
logger.warning(
"Invalid logprobs data. Expected 'logprobs' to contain 'token_logprobs' list."
)
else:
logger.error(
f"Invalid response for loglikelihood. Response: {response}"
)
assert False
return res
def generate_until(self, requests):
if not requests:
return []
res = []
for request in tqdm([req.args for req in requests]):
inp = request[0]
request_args = request[1]
until = request_args.get("until", ["</s>"])
response = self.gguf_completion(context=inp, stop=until)
if response and "choices" in response and response["choices"]:
choice = response["choices"][0]
if "text" in choice:
generated_text = choice["text"].strip()
res.append(generated_text)
else:
logger.error(
f"Invalid response for greedy_until. Response: {response}"
)
res.append(None) # Add default value in case of error
else:
logger.error(f"Invalid response for greedy_until. Response: {response}")
res.append(None) # Add default value in case of error
return res
def loglikelihood_rolling(self, requests):
raise NotImplementedError(
"loglikelihood_rolling not yet supported for GGUF models"
)
lm_eval/models/huggingface.py
View file @ 5d3bf2e7
...@@ -16,13 +16,14 @@ from pathlib import Path ...@@ -16,13 +16,14 @@ from pathlib import Path
import torch.nn.functional as F import torch.nn.functional as F
from lm_eval import utils from lm_eval import utils
from lm_eval.api.instance import Instance
from lm_eval.api.model import LM from lm_eval.api.model import LM
from lm_eval.api.registry import register_model from lm_eval.api.registry import register_model
from lm_eval.utils import MultiTokenEOSCriteria, stop_sequences_criteria from lm_eval.utils import MultiTokenEOSCriteria, stop_sequences_criteria
from accelerate import Accelerator, find_executable_batch_size, DistributedType from accelerate import Accelerator, find_executable_batch_size, DistributedType
from typing import List, Optional, Union from typing import List, Optional, Union, Tuple
eval_logger = utils.eval_logger eval_logger = utils.eval_logger
...@@ -420,7 +421,9 @@ class HFLM(LM): ...@@ -420,7 +421,9 @@ class HFLM(LM):
utils.clear_torch_cache() utils.clear_torch_cache()
return batch_size return batch_size
def tok_encode(self, string: str, left_truncate_len=None, add_special_tokens=None): def tok_encode(
self, string: str, left_truncate_len=None, add_special_tokens=None
) -> List[int]:
""" """ """ """
if add_special_tokens is None: if add_special_tokens is None:
if self.AUTO_MODEL_CLASS == transformers.AutoModelForCausalLM: if self.AUTO_MODEL_CLASS == transformers.AutoModelForCausalLM:
...@@ -442,7 +445,7 @@ class HFLM(LM): ...@@ -442,7 +445,7 @@ class HFLM(LM):
padding_side: str = "left", padding_side: str = "left",
left_truncate_len: int = None, left_truncate_len: int = None,
truncation: bool = False, truncation: bool = False,
): ) -> Tuple[List[int], List[int]]:
# encode a batch of strings. converts to tensors and pads automatically, unlike tok_encode. # encode a batch of strings. converts to tensors and pads automatically, unlike tok_encode.
old_padding_side = self.tokenizer.padding_side old_padding_side = self.tokenizer.padding_side
self.tokenizer.padding_side = padding_side self.tokenizer.padding_side = padding_side
...@@ -536,7 +539,9 @@ class HFLM(LM): ...@@ -536,7 +539,9 @@ class HFLM(LM):
return logits return logits
def _encode_pair(self, context, continuation): def _encode_pair(
self, context: str, continuation: str
) -> Tuple[List[int], List[int]]:
n_spaces = len(context) - len(context.rstrip()) n_spaces = len(context) - len(context.rstrip())
if n_spaces > 0: if n_spaces > 0:
continuation = context[-n_spaces:] + continuation continuation = context[-n_spaces:] + continuation
...@@ -551,7 +556,7 @@ class HFLM(LM): ...@@ -551,7 +556,7 @@ class HFLM(LM):
continuation_enc = whole_enc[context_enc_len:] continuation_enc = whole_enc[context_enc_len:]
return context_enc, continuation_enc return context_enc, continuation_enc
def loglikelihood(self, requests): def loglikelihood(self, requests: List[Instance]) -> List[Tuple[float, bool]]:
new_reqs = [] new_reqs = []
for context, continuation in [req.args for req in requests]: for context, continuation in [req.args for req in requests]:
if context == "": if context == "":
...@@ -566,7 +571,7 @@ class HFLM(LM): ...@@ -566,7 +571,7 @@ class HFLM(LM):
return self._loglikelihood_tokens(new_reqs) return self._loglikelihood_tokens(new_reqs)
def loglikelihood_rolling(self, requests): def loglikelihood_rolling(self, requests: List[Instance]) -> List[float]:
loglikelihoods = [] loglikelihoods = []
adaptive_batch_size = None adaptive_batch_size = None
...@@ -640,8 +645,11 @@ class HFLM(LM): ...@@ -640,8 +645,11 @@ class HFLM(LM):
return self.batch_sizes[sched] return self.batch_sizes[sched]
def _loglikelihood_tokens( def _loglikelihood_tokens(
self, requests, disable_tqdm: bool = False, override_bs=None self,
): requests: List[Tuple[Tuple[str, str], List[int], List[int]]],
disable_tqdm: bool = False,
override_bs: int = None,
) -> List[Tuple[float, bool]]:
# TODO: implement some kind of efficient-request-middleware that lumps together requests with the same context # TODO: implement some kind of efficient-request-middleware that lumps together requests with the same context
res = [] res = []
...@@ -820,7 +828,7 @@ class HFLM(LM): ...@@ -820,7 +828,7 @@ class HFLM(LM):
return re_ord.get_original(res) return re_ord.get_original(res)
def generate_until(self, requests): def generate_until(self, requests: List[Instance]) -> List[str]:
res = defaultdict(list) res = defaultdict(list)
re_ords = {} re_ords = {}
...@@ -889,8 +897,6 @@ class HFLM(LM): ...@@ -889,8 +897,6 @@ class HFLM(LM):
max_gen_toks = kwargs.pop("max_gen_toks") max_gen_toks = kwargs.pop("max_gen_toks")
else: else:
max_gen_toks = self.max_gen_toks max_gen_toks = self.max_gen_toks
# first stop sequence is used to halt generation upon encountering
primary_until = [until[0]]
# set the max length in tokens of inputs ("context_enc") # set the max length in tokens of inputs ("context_enc")
if self.AUTO_MODEL_CLASS == transformers.AutoModelForCausalLM: if self.AUTO_MODEL_CLASS == transformers.AutoModelForCausalLM:
...@@ -916,7 +922,7 @@ class HFLM(LM): ...@@ -916,7 +922,7 @@ class HFLM(LM):
cont = self._model_generate( cont = self._model_generate(
context=context_enc, context=context_enc,
attention_mask=attn_masks, attention_mask=attn_masks,
stop=primary_until, stop=until,
**kwargs, **kwargs,
) )
......
lm_eval/models/vllm_causallms.py 0 → 100644
View file @ 5d3bf2e7
from collections import defaultdict
from typing import List, Tuple, Optional, Literal, Union
from lm_eval.api.instance import Instance
from lm_eval.api.model import LM
import copy
from tqdm import tqdm
from lm_eval.api.registry import register_model
from lm_eval import utils
try:
from vllm import LLM, SamplingParams
except ModuleNotFoundError:
pass
eval_logger = utils.eval_logger
@register_model("vllm")
class VLLM(LM):
_DEFAULT_MAX_LENGTH = 2048
def __init__(
self,
pretrained="gpt2",
dtype: Literal["float16", "bfloat16", "float32", "auto"] = "auto",
revision: Optional[str] = None,
trust_remote_code: Optional[bool] = False,
tokenizer_mode: Literal["auto", "slow"] = "auto",
tensor_parallel_size: int = 1,
quantization: Optional[Literal["awq"]] = None,
max_gen_toks: int = 256,
swap_space: int = 4,
batch_size: Union[str, int] = 1,
max_batch_size=None,
max_length: int = None,
seed: int = 1234,
gpu_memory_utilization: float = 0.9,
device: str = "cuda",
):
super().__init__()
try:
import vllm
except ModuleNotFoundError:
raise Exception(
"attempted to use 'vllm' LM type, but package `vllm` is not installed. \
please install vllm via `pip install lm-eval[vllm]` or `pip install -e .[vllm]`",
)
assert "cuda" in device or device is None, "vLLM only supports CUDA"
self.model = LLM(
model=pretrained,
gpu_memory_utilization=float(gpu_memory_utilization),
revision=revision,
dtype=dtype,
tokenizer_mode=tokenizer_mode,
trust_remote_code=trust_remote_code,
tensor_parallel_size=int(tensor_parallel_size),
swap_space=int(swap_space),
quantization=quantization,
seed=int(seed),
)
self.tokenizer = self.model.get_tokenizer()
self.batch_size = batch_size
self._max_length = max_length
self._max_gen_toks = max_gen_toks
@property
def eot_token_id(self):
# we use EOT because end of *text* is more accurate for what we're doing than end of *sentence*
return self.tokenizer.eos_token_id
@property
def max_length(self):
if self._max_length: # if max length manually set, return it
return self._max_length
if hasattr(self.model.llm_engine.model_config, "max_model_len"):
return self.model.llm_engine.model_config.max_model_len
return self._DEFAULT_MAX_LENGTH
@property
def max_gen_toks(self):
return self._max_gen_toks
def tok_encode(
self,
string: str,
left_truncate_len=None,
add_special_tokens=False,
truncation=False,
):
""" """
encoding = self.tokenizer.encode(
string, add_special_tokens=add_special_tokens, truncation=truncation
)
# left-truncate the encoded context to be at most `left_truncate_len` tokens long
if left_truncate_len:
encoding = encoding[-left_truncate_len:]
return encoding
def _model_generate(
self,
requests: List[int] = None,
generate: bool = False,
max_tokens: int = None,
stop: Optional[List[str]] = None,
use_tqdm=True,
**kwargs,
):
if "do_sample" in kwargs.keys():
kwargs.pop("do_sample")
if generate:
generate_sampling_params = SamplingParams(
max_tokens=max_tokens, stop=stop, **kwargs
)
outputs = self.model.generate(
prompt_token_ids=requests,
sampling_params=generate_sampling_params,
use_tqdm=use_tqdm,
)
else:
logliklihood_sampling_params = SamplingParams(
temperature=0, prompt_logprobs=2, max_tokens=1
)
outputs = self.model.generate(
prompt_token_ids=requests,
sampling_params=logliklihood_sampling_params,
use_tqdm=use_tqdm,
)
return outputs
def loglikelihood(self, requests: List[Instance]) -> List[Tuple[float, bool]]:
new_reqs = []
for context, continuation in [req.args for req in requests]:
if context == "":
# end of text as context
context_enc, continuation_enc = [self.eot_token_id], self.tok_encode(
continuation
)
else:
context_enc, continuation_enc = self.tokenizer(
[context, continuation],
truncation="do_not_truncate",
add_special_tokens=False,
return_attention_mask=False,
).input_ids
new_reqs.append(((context, continuation), context_enc, continuation_enc))
return self._loglikelihood_tokens(new_reqs)
def loglikelihood_rolling(self, requests: List[Instance]) -> List[float]:
loglikelihoods = []
for (string,) in tqdm([req.args for req in requests]):
rolling_token_windows = list(
map(
utils.make_disjoint_window,
utils.get_rolling_token_windows(
token_list=self.tok_encode(string),
prefix_token=self.eot_token_id,
max_seq_len=self.max_length - 1,
context_len=1,
),
)
)
rolling_token_windows = [(None,) + x for x in rolling_token_windows]
string_nll = self._loglikelihood_tokens(
rolling_token_windows,
)
# discard is_greedy
string_nll = [x[0] for x in string_nll]
string_nll = sum(string_nll)
loglikelihoods.append(string_nll)
return loglikelihoods
def generate_until(self, requests: List[Instance]) -> List[str]:
res = defaultdict(list)
re_ords = {}
# batch tokenize contexts
context, all_gen_kwargs = zip(*(req.args for req in requests))
context_encoding = self.tokenizer(context).input_ids
requests = [
((a, b), c) for a, b, c in zip(context, context_encoding, all_gen_kwargs)
]
def _collate_gen(_requests):
# the negative sign on len(toks) sorts descending - this has a few advantages:
# - time estimates will always be over not underestimates, which is more useful for planning
# - to know the size of a batch when going through the list, you know the first one is always the batch
# padded context length. this is useful to simplify the batching logic and more importantly to make
# automatic adaptive batches much much easier to implement
# - any OOMs will happen right away rather than near the end
return -len(_requests[0][1]), tuple(_requests[0][1])
# we group requests by their generation_kwargs,
# so that we don't try to execute e.g. greedy sampling and temp=0.8 sampling
# in the same batch.
grouper = utils.Grouper(requests, lambda x: str(x[1]))
for key, reqs in grouper.get_grouped().items():
# within each set of reqs for given kwargs, we reorder by token length, descending.
re_ords[key] = utils.Reorderer(requests, _collate_gen)
pbar = tqdm(total=len(requests), disable=(self.rank != 0))
# for each different set of kwargs, we execute all requests, by batch.
for key, re_ord in re_ords.items():
chunks = utils.chunks(
re_ord.get_reordered(),
n=self.batch_size if self.batch_size != "auto" else 0,
fn=None,
)
for chunk in chunks:
context_and_encoding, all_gen_kwargs = zip(*chunk)
context, context_encoding = zip(*context_and_encoding)
# we assume all gen kwargs in the batch are the same
# this is safe to assume because the `grouper` object ensures it.
gen_kwargs = all_gen_kwargs[0]
# unpack our keyword arguments.
until = None
if isinstance(gen_kwargs, dict):
kwargs = copy.deepcopy(gen_kwargs) # edge case for repeats > 1
if "until" in kwargs.keys():
until = kwargs.pop("until")
if isinstance(until, str):
until = [until]
elif not isinstance(until, list):
raise ValueError(
f"Expected `kwargs['until']` to be of type Union[str,list] but got {until}"
)
else:
raise ValueError(
f"Expected `kwargs` to be of type `dict` but got {gen_kwargs}"
)
if not until:
until = [self.tokenizer.decode(self.eot_token_id)]
if "max_gen_toks" in kwargs.keys():
max_gen_toks = kwargs.pop("max_gen_toks")
else:
max_gen_toks = self.max_gen_toks
# set the max length in tokens of inputs ("context_enc")
# max len for inputs = max length, minus room to generate the max new tokens
max_ctx_len = self.max_length - max_gen_toks
context_encoding = [x[-max_ctx_len:] for x in context_encoding]
# TODO: max_length in kwargs
# perform batched generation
cont = self._model_generate(
requests=context_encoding,
generate=True,
max_tokens=max_gen_toks,
stop=until,
**kwargs,
)
# cache generations
for output, context in zip(cont, context):
generated_text = output.outputs[0].text
res[key].append(generated_text)
self.cache_hook.add_partial(
"generate_until", (context, gen_kwargs), generated_text
)
pbar.update(1)
# reorder this group of results back to original unsorted form
res[key] = re_ord.get_original(res[key])
pbar.close()
return grouper.get_original(res)
def _loglikelihood_tokens(
self,
requests: List[Tuple[Tuple[str, str], List[int], List[int]]],
disable_tqdm: bool = False,
) -> List[Tuple[float, bool]]:
res = []
def _collate(x):
toks = x[1] + x[2]
return -len(toks), tuple(toks)
re_ord = utils.Reorderer(requests, _collate)
chunks = utils.chunks(
re_ord.get_reordered(),
n=self.batch_size if self.batch_size != "auto" else 0,
fn=None,
)
pbar = tqdm(total=len(requests), disable=disable_tqdm)
for chunk in chunks:
inps = []
ctxlens = []
for cache_key, context_enc, continuation_enc in chunk:
inp = (context_enc + continuation_enc)[-(self.max_length) :]
ctxlen = len(context_enc) - max(
0, len(context_enc) + len(continuation_enc) - (self.max_length)
)
inps.append(inp)
ctxlens.append(ctxlen)
outputs = self._model_generate(requests=inps, generate=False)
for output, ctxlen, (cache_key, context_enc, continuation_enc) in zip(
outputs, ctxlens, chunk
):
answer = self._parse_logprobs(
(context_enc + continuation_enc),
output,
ctxlen,
)
res.append(answer)
# partial caching
if cache_key is not None:
self.cache_hook.add_partial("loglikelihood", cache_key, answer)
pbar.update(1)
pbar.close()
return re_ord.get_original(res)
@staticmethod
def _parse_logprobs(tokens: List, outputs, ctxlen: int) -> Tuple[float, bool]:
"""Process logprobs and tokens.
:param tokens: list
Tokens from context+continuations
:param outputs: RequestOutput
Contains prompt
:param ctxlen: int
Length of context (so we can slice them away and only keep the predictions)
:return:
continuation_logprobs: float
Log probabilities of continuation tokens
is_greedy: bool
Whether argmax matches given continuation exactly
"""
# prompt_logprobs = [None, {}*len(context-1)]
continuation_logprobs_dicts = outputs.prompt_logprobs
# Calculate continuation_logprobs
# assume ctxlen always > 1
continuation_logprobs = sum(
logprob_dict.get(token)
for token, logprob_dict in zip(
tokens[ctxlen:], continuation_logprobs_dicts[ctxlen:]
)
)
# Determine if is_greedy
is_greedy = True
for token, logprob_dict in zip(
tokens[ctxlen:], continuation_logprobs_dicts[ctxlen:]
):
# Get the token with the maximum log probability from the logprob_dict
if logprob_dict: # Ensure the logprob_dict is not None
top_token = max(logprob_dict, key=logprob_dict.get)
if top_token != token:
is_greedy = False
break
return continuation_logprobs, is_greedy
lm_eval/tasks/bbh/README.md
View file @ 5d3bf2e7
...@@ -25,10 +25,10 @@ Homepage: https://github.com/suzgunmirac/BIG-Bench-Hard ...@@ -25,10 +25,10 @@ Homepage: https://github.com/suzgunmirac/BIG-Bench-Hard
#### Groups #### Groups
- `bbh_flan_zeroshot` - `bbh_zeroshot`
- `bbh_flan_fewshot` - `bbh_fewshot`
- `bbh_flan_cot_fewshot` - `bbh_cot_fewshot`
- `bbh_flan_cot_zeroshot` - `bbh_cot_zeroshot`
#### Tasks #### Tasks
......
lm_eval/tasks/bbh/_generate_configs.py
View file @ 5d3bf2e7
...@@ -17,7 +17,7 @@ from lm_eval.logger import eval_logger ...@@ -17,7 +17,7 @@ from lm_eval.logger import eval_logger
def parse_args(): def parse_args():
parser = argparse.ArgumentParser() parser = argparse.ArgumentParser()
parser.add_argument("--base_yaml_path", required=True) parser.add_argument("--base_yaml_path", required=True)
parser.add_argument("--save_prefix_path", default="flan_zeroshot") parser.add_argument("--save_prefix_path", default="zeroshot")
parser.add_argument("--cot", default=False) parser.add_argument("--cot", default=False)
parser.add_argument("--fewshot", default=False) parser.add_argument("--fewshot", default=False)
parser.add_argument("--task_prefix", default="") parser.add_argument("--task_prefix", default="")
......
lm_eval/tasks/bbh/flan_cot_fewshot/_flan_cot_fewshot_template_yaml → lm_eval/tasks/bbh/cot_fewshot/_cot_fewshot_template_yaml
View file @ 5d3bf2e7
group: bbh_flan_cot_fewshot group:
- bbh
- bbh_cot_fewshot
dataset_path: lukaemon/bbh dataset_path: lukaemon/bbh
output_type: generate_until output_type: generate_until
test_split: test test_split: test
...@@ -12,6 +14,8 @@ metric_list: ...@@ -12,6 +14,8 @@ metric_list:
generation_kwargs: generation_kwargs:
until: until:
- "</s>" - "</s>"
- "Q"
- "\n\n"
do_sample: false do_sample: false
temperature: 0.0 temperature: 0.0
filter_list: filter_list:
......
lm_eval/tasks/bbh/flan_cot_fewshot/boolean_expressions.yaml → lm_eval/tasks/bbh/cot_fewshot/boolean_expressions.yaml
View file @ 5d3bf2e7
"dataset_name": "boolean_expressions" "dataset_name": "boolean_expressions"
"description": "Evaluate the result of a random Boolean expression.\n\n" "description": "Evaluate the result of a random Boolean expression.\n\n"
"doc_to_text": " not ( ( not not True ) ) is\nA: Let's think step by step.\nRemember that (i) expressions inside brackets are always evaluated first and that (ii) the order of operations from highest priority to lowest priority is \"not\", \"and\", \"or\", respectively.\nWe first simplify this expression \"Z\" as follows: \"Z = not ( ( not not True ) ) = not ( ( A ) )\" where \"A = not not True\".\nLet's evaluate A: A = not not True = not (not True) = not False = True.\nPlugging in A, we get: Z = not ( ( A ) ) = not ( ( True ) ) = not True = False. So the answer is False. True and False and not True and True is\nA: Let's think step by step.\nRemember that (i) expressions inside brackets are always evaluated first and that (ii) the order of operations from highest priority to lowest priority is \"not\", \"and\", \"or\", respectively.\nWe first simplify this expression \"Z\" as follows: \"Z = True and False and not True and True = A and B\" where \"A = True and False\" and \"B = not True and True\".\nLet's evaluate A: A = True and False = False.\nLet's evaluate B: B = not True and True = not (True and True) = not (True) = False.\nPlugging in A and B, we get: Z = A and B = False and False = False. So the answer is False. not not ( not ( False ) ) is\nA: Let's think step by step.\nRemember that (i) expressions inside brackets are always evaluated first and that (ii) the order of operations from highest priority to lowest priority is \"not\", \"and\", \"or\", respectively.\nWe first simplify this expression \"Z\" as follows: \"Z = not not ( not ( False ) ) = not not ( A )\" where \"A = not ( False )\".\nLet's evaluate A: A = not ( False ) = not False = True.\nPlugging in A, we get: Z = not not ( A ) = not not (True) = not not False = True. So the answer is True.Q: {{input}}\nA: Let's think step by step.\n" "doc_to_text": " not ( ( not not True ) ) is\nA: Let's think step by step.\nRemember that (i) expressions inside brackets are always evaluated first and that (ii) the order of operations from highest priority to lowest priority is \"not\", \"and\", \"or\", respectively.\nWe first simplify this expression \"Z\" as follows: \"Z = not ( ( not not True ) ) = not ( ( A ) )\" where \"A = not not True\".\nLet's evaluate A: A = not not True = not (not True) = not False = True.\nPlugging in A, we get: Z = not ( ( A ) ) = not ( ( True ) ) = not True = False. So the answer is False. True and False and not True and True is\nA: Let's think step by step.\nRemember that (i) expressions inside brackets are always evaluated first and that (ii) the order of operations from highest priority to lowest priority is \"not\", \"and\", \"or\", respectively.\nWe first simplify this expression \"Z\" as follows: \"Z = True and False and not True and True = A and B\" where \"A = True and False\" and \"B = not True and True\".\nLet's evaluate A: A = True and False = False.\nLet's evaluate B: B = not True and True = not (True and True) = not (True) = False.\nPlugging in A and B, we get: Z = A and B = False and False = False. So the answer is False. not not ( not ( False ) ) is\nA: Let's think step by step.\nRemember that (i) expressions inside brackets are always evaluated first and that (ii) the order of operations from highest priority to lowest priority is \"not\", \"and\", \"or\", respectively.\nWe first simplify this expression \"Z\" as follows: \"Z = not not ( not ( False ) ) = not not ( A )\" where \"A = not ( False )\".\nLet's evaluate A: A = not ( False ) = not False = True.\nPlugging in A, we get: Z = not not ( A ) = not not (True) = not not False = True. So the answer is True.Q: {{input}}\nA: Let's think step by step.\n"
"include": "_flan_cot_fewshot_template_yaml" "include": "_cot_fewshot_template_yaml"
"task": "bbh_flan_cot_fewshot_boolean_expressions" "task": "bbh_cot_fewshot_boolean_expressions"
lm_eval/tasks/bbh/flan_cot_fewshot/causal_judgement.yaml → lm_eval/tasks/bbh/cot_fewshot/causal_judgement.yaml
View file @ 5d3bf2e7
"dataset_name": "causal_judgement" "dataset_name": "causal_judgement"
"description": "Answer questions about causal attribution.\n\n" "description": "Answer questions about causal attribution.\n\n"
"doc_to_text": " How would a typical person answer each of the following questions about causation?\nFrank T., had an ongoing dispute with his neighbor over a stretch of land and one day decided to shoot his neighbor in the body. Frank T. had no experience with guns, his hand slipped on the barrel of the gun, and the shot went wild. Nonetheless, the bullet bounced off a large boulder several feet away and hit the neighbor's body, causing significant injury. Did Frank T. intentionally shoot his neighbor in the body?\nOptions:\n- Yes\n- No\nA: Let's think step by step.\nHere in this question, we are told that \"Frank T. had no experience with guns, his hand slipped on the barrel of the gun, and the shot went wild.\" A typical person would assume that this passage suggests that Frank T. had no intention of shooting and injuring someone and that the bullet accidentally hit the neighbor's body; therefore, we conclude that Frank T. did not intentionally hit his neighbor. So the answer is No. How would a typical person answer each of the following questions about causation?\nSuzy and Billy are working on a project that is very important for our nation's security. The boss tells them both: \"Be sure that you are here at exactly 9 am. It is absolutely essential that you arrive at that time.\" Both Billy and Suzy arrive at 9 am. As it happens, there was a motion detector installed in the room where they arrived. The motion detector was set up to be triggered if at least one person appeared in the room at the same time. So the motion detector went off. Did Billy cause the motion detector to go off?\nOptions:\n- Yes\n- No\nA: Let's think step by step.\nHere in this question, we are told that the boss ordered them both to arrive at the meeting room at the same time and that the motion detector was set up to be triggered if at least one person appeared in the room at the same time.\" A typical person would assume that the person probably meant to say the detector was set up to be triggered if \"both persons\" appeared in the room at the same time, not at least one person, since otherwise the phrase \"at the same time\" would not make much sense in that sentence. Because the motion detector went off, a typical person would therefore come to the conclusion that both Suzy and Billy triggered the motion detector to go off; hence, Billy did indeed cause the motion detector to go off. So the answer is Yes. How would a typical person answer each of the following questions about causation?\nGeorge and his sister Lena reunite at their parents' house for Thanksgiving. Whereas George just got into medical school, Lena is unhappy in her marriage and recently lost her job. Over the course of the day, George and Lena get into a number of heated arguments. Later in the afternoon they play a game of darts. They split the first two games, and the third game is close until the end. Who will win comes down to George's last shot. If he hits a high point region, he wins; if he hits a low point region, Lena wins. George thinks of the difficult time Lena is having, and he really wants to let her win. He aims the dart at the low point region. He sets up his shot and the dart lands in the low point region. After his shot, Lena wins the game and is very happy. Did George hit the low point region intentionally?\nOptions:\n- Yes\n- No\nA: Let's think step by step.\nHere in this question, we are told that \"He aims the dart at the low point region.\" A typical person might therefore think George did intentionally hit the low point region, because he wanted to lift up the spirit of his sister Lena. So the answer is Yes.Q: {{input}}\nA: Let's think step by step.\n" "doc_to_text": " How would a typical person answer each of the following questions about causation?\nFrank T., had an ongoing dispute with his neighbor over a stretch of land and one day decided to shoot his neighbor in the body. Frank T. had no experience with guns, his hand slipped on the barrel of the gun, and the shot went wild. Nonetheless, the bullet bounced off a large boulder several feet away and hit the neighbor's body, causing significant injury. Did Frank T. intentionally shoot his neighbor in the body?\nOptions:\n- Yes\n- No\nA: Let's think step by step.\nHere in this question, we are told that \"Frank T. had no experience with guns, his hand slipped on the barrel of the gun, and the shot went wild.\" A typical person would assume that this passage suggests that Frank T. had no intention of shooting and injuring someone and that the bullet accidentally hit the neighbor's body; therefore, we conclude that Frank T. did not intentionally hit his neighbor. So the answer is No. How would a typical person answer each of the following questions about causation?\nSuzy and Billy are working on a project that is very important for our nation's security. The boss tells them both: \"Be sure that you are here at exactly 9 am. It is absolutely essential that you arrive at that time.\" Both Billy and Suzy arrive at 9 am. As it happens, there was a motion detector installed in the room where they arrived. The motion detector was set up to be triggered if at least one person appeared in the room at the same time. So the motion detector went off. Did Billy cause the motion detector to go off?\nOptions:\n- Yes\n- No\nA: Let's think step by step.\nHere in this question, we are told that the boss ordered them both to arrive at the meeting room at the same time and that the motion detector was set up to be triggered if at least one person appeared in the room at the same time.\" A typical person would assume that the person probably meant to say the detector was set up to be triggered if \"both persons\" appeared in the room at the same time, not at least one person, since otherwise the phrase \"at the same time\" would not make much sense in that sentence. Because the motion detector went off, a typical person would therefore come to the conclusion that both Suzy and Billy triggered the motion detector to go off; hence, Billy did indeed cause the motion detector to go off. So the answer is Yes. How would a typical person answer each of the following questions about causation?\nGeorge and his sister Lena reunite at their parents' house for Thanksgiving. Whereas George just got into medical school, Lena is unhappy in her marriage and recently lost her job. Over the course of the day, George and Lena get into a number of heated arguments. Later in the afternoon they play a game of darts. They split the first two games, and the third game is close until the end. Who will win comes down to George's last shot. If he hits a high point region, he wins; if he hits a low point region, Lena wins. George thinks of the difficult time Lena is having, and he really wants to let her win. He aims the dart at the low point region. He sets up his shot and the dart lands in the low point region. After his shot, Lena wins the game and is very happy. Did George hit the low point region intentionally?\nOptions:\n- Yes\n- No\nA: Let's think step by step.\nHere in this question, we are told that \"He aims the dart at the low point region.\" A typical person might therefore think George did intentionally hit the low point region, because he wanted to lift up the spirit of his sister Lena. So the answer is Yes.Q: {{input}}\nA: Let's think step by step.\n"
"include": "_flan_cot_fewshot_template_yaml" "include": "_cot_fewshot_template_yaml"
"task": "bbh_flan_cot_fewshot_causal_judgement" "task": "bbh_cot_fewshot_causal_judgement"
lm_eval/tasks/bbh/flan_cot_fewshot/date_understanding.yaml → lm_eval/tasks/bbh/cot_fewshot/date_understanding.yaml
View file @ 5d3bf2e7
"dataset_name": "date_understanding" "dataset_name": "date_understanding"
"description": "Infer the date from context.\n\n" "description": "Infer the date from context.\n\n"
"doc_to_text": " Today is Christmas Eve of 1937. What is the date 10 days ago in MM/DD/YYYY?\nOptions:\n(A) 12/14/2026\n(B) 12/14/1950\n(C) 12/14/2007\n(D) 12/14/1937\n(E) 07/14/1938\n(F) 12/14/1988\nA: Let's think step by step.\nIf today is Christmas Eve of 1937, then today's date is December 24, 1937. 10 days before today is December 14, 1937, that is 12/14/1937. So the answer is (D). Tomorrow is 11/12/2019. What is the date one year ago from today in MM/DD/YYYY?\nOptions:\n(A) 09/04/2018\n(B) 11/11/2018\n(C) 08/25/2018\n(D) 11/02/2018\n(E) 11/04/2018\nA: Let's think step by step.\nIf tomorrow is 11/12/2019, then today is 11/11/2019. The date one year ago from today is 11/11/2018. So the answer is (B). Jane and John married on Jan 2, 1958. It is their 5-year anniversary today. What is the date tomorrow in MM/DD/YYYY?\nOptions:\n(A) 01/11/1961\n(B) 01/03/1963\n(C) 01/18/1961\n(D) 10/14/1960\n(E) 01/03/1982\n(F) 12/03/1960\nA: Let's think step by step.\nIf Jane and John married on Jan 2, 1958, then and if it is their 5-year anniversary today, then today's date is Jan 2, 1963. The date tomorrow is Jan 3, 1963, that is 01/03/1963. So the answer is (B).Q: {{input}}\nA: Let's think step by step.\n" "doc_to_text": " Today is Christmas Eve of 1937. What is the date 10 days ago in MM/DD/YYYY?\nOptions:\n(A) 12/14/2026\n(B) 12/14/1950\n(C) 12/14/2007\n(D) 12/14/1937\n(E) 07/14/1938\n(F) 12/14/1988\nA: Let's think step by step.\nIf today is Christmas Eve of 1937, then today's date is December 24, 1937. 10 days before today is December 14, 1937, that is 12/14/1937. So the answer is (D). Tomorrow is 11/12/2019. What is the date one year ago from today in MM/DD/YYYY?\nOptions:\n(A) 09/04/2018\n(B) 11/11/2018\n(C) 08/25/2018\n(D) 11/02/2018\n(E) 11/04/2018\nA: Let's think step by step.\nIf tomorrow is 11/12/2019, then today is 11/11/2019. The date one year ago from today is 11/11/2018. So the answer is (B). Jane and John married on Jan 2, 1958. It is their 5-year anniversary today. What is the date tomorrow in MM/DD/YYYY?\nOptions:\n(A) 01/11/1961\n(B) 01/03/1963\n(C) 01/18/1961\n(D) 10/14/1960\n(E) 01/03/1982\n(F) 12/03/1960\nA: Let's think step by step.\nIf Jane and John married on Jan 2, 1958, then and if it is their 5-year anniversary today, then today's date is Jan 2, 1963. The date tomorrow is Jan 3, 1963, that is 01/03/1963. So the answer is (B).Q: {{input}}\nA: Let's think step by step.\n"
"include": "_flan_cot_fewshot_template_yaml" "include": "_cot_fewshot_template_yaml"
"task": "bbh_flan_cot_fewshot_date_understanding" "task": "bbh_cot_fewshot_date_understanding"
lm_eval/tasks/bbh/flan_cot_fewshot/disambiguation_qa.yaml → lm_eval/tasks/bbh/cot_fewshot/disambiguation_qa.yaml
View file @ 5d3bf2e7
"dataset_name": "disambiguation_qa" "dataset_name": "disambiguation_qa"
"description": "Clarify the meaning of sentences with ambiguous pronouns.\n\n" "description": "Clarify the meaning of sentences with ambiguous pronouns.\n\n"
"doc_to_text": " In the following sentences, explain the antecedent of the pronoun (which thing the pronoun refers to), or state that it is ambiguous.\nSentence: The chief told the counselor that they took the day off.\nOptions:\n(A) The chief took the day off\n(B) The counselor took the day off\n(C) Ambiguous\nA: Let's think step by step.\nHere we need to determine who the pronoun \"they\" might be referring to. There are two possible referents for \"they\", namely the chief and the counselor. The verb \"told\" might be able to help us determine which one is more likely (if either). Let X be the chief and Y the counselor. The sentence is then of the form \"X told Y that (X or Y) did something.\"\nLet's consider Y first: \"X told Y that Y did something.\" This case does not make much sense, as Y would already have the information that Y did something, because it is information about themself.\nNow, consider X: \"X told Y that X did something.\" This makes sense, because X would be sharing some information about themself that Y might not have known before.\nBecause in this context, X is the chief and Y is the counselor, the answer should be the chief. So the answer is (A). In the following sentences, explain the antecedent of the pronoun (which thing the pronoun refers to), or state that it is ambiguous.\nSentence: The manager sent a message to the secretary, but he didn't reply yet.\nOptions:\n(A) The secretary didn't reply yet\n(B) The manager didn't reply yet\n(C) Ambiguous\nA: Let's think step by step.\nHere we need to determine who the pronoun \"he\" might be referring to. There are two possible referents for \"he\", namely the manager and the secretary. The verbs \"sent\" and \"reply\" might be able to help us determine which one is more likely (if either). Let X be the manager and Y the secretary. The sentence is then of the form \"X sent a message to Y, but (X or Y) didn't reply yet.\"\nLet's consider Y first: \"X sent a message to Y, but Y didn't reply yet.\" This case makes sense, because of the implicit causality of the sentence. Y was the receiver of the message, but Y didn't get back to X yet.\nNow, consider X: \"X sent a message to Y, but X didn't reply yet.\" This case doesn't make sense, because X was the initial sender of the message, so it is now Y's turn to write back to X.\nBecause in this context, X is the manager and Y is the secretary, the answer should be the secretary. So the answer is (A). In the following sentences, explain the antecedent of the pronoun (which thing the pronoun refers to), or state that it is ambiguous.\nSentence: Bailey will plan to meet the director at his office\nOptions:\n(A) It will be Bailey's office\n(B) It will be the director's office\n(C) Ambiguous\nA: Let's think step by step.\nHere we need to determine who the pronoun \"his\" might be referring to. There are two possible referents for \"his\", namely Bailey's and the director's. The verb phrase \"plan to meet\" might be able to help us determine which one is more likely (if either). Let X be Bailey and Y the director. The sentence is then of the form \"X will plan to meet Y at (X or Y)'s office.\"\nLet's consider Y first: \"X will plan to meet Y at Y's office.\" This case makes sense, because X might want to meet up with Y at Y's office.\nNow, consider X: \"X will plan to meet Y at X's office.\" This case also makes sense, because X might want to meet up with Y at X's own office.\nBecause both X and Y are possible at the same time, we conclude that the antecedent of the pronoun is ambiguous. So the answer is (C).Q: {{input}}\nA: Let's think step by step.\n" "doc_to_text": " In the following sentences, explain the antecedent of the pronoun (which thing the pronoun refers to), or state that it is ambiguous.\nSentence: The chief told the counselor that they took the day off.\nOptions:\n(A) The chief took the day off\n(B) The counselor took the day off\n(C) Ambiguous\nA: Let's think step by step.\nHere we need to determine who the pronoun \"they\" might be referring to. There are two possible referents for \"they\", namely the chief and the counselor. The verb \"told\" might be able to help us determine which one is more likely (if either). Let X be the chief and Y the counselor. The sentence is then of the form \"X told Y that (X or Y) did something.\"\nLet's consider Y first: \"X told Y that Y did something.\" This case does not make much sense, as Y would already have the information that Y did something, because it is information about themself.\nNow, consider X: \"X told Y that X did something.\" This makes sense, because X would be sharing some information about themself that Y might not have known before.\nBecause in this context, X is the chief and Y is the counselor, the answer should be the chief. So the answer is (A). In the following sentences, explain the antecedent of the pronoun (which thing the pronoun refers to), or state that it is ambiguous.\nSentence: The manager sent a message to the secretary, but he didn't reply yet.\nOptions:\n(A) The secretary didn't reply yet\n(B) The manager didn't reply yet\n(C) Ambiguous\nA: Let's think step by step.\nHere we need to determine who the pronoun \"he\" might be referring to. There are two possible referents for \"he\", namely the manager and the secretary. The verbs \"sent\" and \"reply\" might be able to help us determine which one is more likely (if either). Let X be the manager and Y the secretary. The sentence is then of the form \"X sent a message to Y, but (X or Y) didn't reply yet.\"\nLet's consider Y first: \"X sent a message to Y, but Y didn't reply yet.\" This case makes sense, because of the implicit causality of the sentence. Y was the receiver of the message, but Y didn't get back to X yet.\nNow, consider X: \"X sent a message to Y, but X didn't reply yet.\" This case doesn't make sense, because X was the initial sender of the message, so it is now Y's turn to write back to X.\nBecause in this context, X is the manager and Y is the secretary, the answer should be the secretary. So the answer is (A). In the following sentences, explain the antecedent of the pronoun (which thing the pronoun refers to), or state that it is ambiguous.\nSentence: Bailey will plan to meet the director at his office\nOptions:\n(A) It will be Bailey's office\n(B) It will be the director's office\n(C) Ambiguous\nA: Let's think step by step.\nHere we need to determine who the pronoun \"his\" might be referring to. There are two possible referents for \"his\", namely Bailey's and the director's. The verb phrase \"plan to meet\" might be able to help us determine which one is more likely (if either). Let X be Bailey and Y the director. The sentence is then of the form \"X will plan to meet Y at (X or Y)'s office.\"\nLet's consider Y first: \"X will plan to meet Y at Y's office.\" This case makes sense, because X might want to meet up with Y at Y's office.\nNow, consider X: \"X will plan to meet Y at X's office.\" This case also makes sense, because X might want to meet up with Y at X's own office.\nBecause both X and Y are possible at the same time, we conclude that the antecedent of the pronoun is ambiguous. So the answer is (C).Q: {{input}}\nA: Let's think step by step.\n"
"include": "_flan_cot_fewshot_template_yaml" "include": "_cot_fewshot_template_yaml"
"task": "bbh_flan_cot_fewshot_disambiguation_qa" "task": "bbh_cot_fewshot_disambiguation_qa"
lm_eval/tasks/bbh/flan_cot_fewshot/dyck_languages.yaml → lm_eval/tasks/bbh/cot_fewshot/dyck_languages.yaml
View file @ 5d3bf2e7
"dataset_name": "dyck_languages" "dataset_name": "dyck_languages"
"description": "Correctly close a Dyck-n word.\n\n" "description": "Correctly close a Dyck-n word.\n\n"
"doc_to_text": " Complete the rest of the sequence, making sure that the parentheses are closed properly. Input: [ { [\nA: Let's think step by step.\nWe should process each input one by one and keep track of the stack configuration.\n0: empty stack\n1: [ ; stack: [\n2: { ; stack: [ {\n3: [ ; stack: [ { [\nNow, we have reached the end. The final stack is \"[ { [\".\nWe will need to pop out \"[\", \"{\", \"[\" one by one in that order.\nSo, we need \"]\", \"}\", \"]\". So the answer is ] } ]. Complete the rest of the sequence, making sure that the parentheses are closed properly. Input: < > ( ( [ [ ( { } ) [ < > ] ]\nA: Let's think step by step.\nWe should process each input one by one and keep track of the stack configuration.\n0: empty stack\n1: < ; stack: <\n2: > ; stack: empty\n3: ( ; stack: (\n4: ( ; stack: ( (\n5: [ ; stack: ( ( [\n6: [ ; stack: ( ( [ [\n7: ( ; stack: ( ( [ [ (\n8: { ; stack: ( ( [ [ ( {\n9: } ; stack: ( ( [ [ (\n10: ) ; stack: ( ( [ [\n11: [ ; stack: ( ( [ [ [\n12: < ; stack: ( ( [ [ [ <\n13: > ; stack: ( ( [ [ [\n14: ] ; stack: ( ( [ [\n15: ] ; stack: ( ( [\nNow, we have reached the end. The final stack is \"( ( [\".\nWe will need to pop out \"[\", \"(\", \"(\" one by one in that order.\nSo, we need \"]\", \")\", \")\". So the answer is ] ) ). Complete the rest of the sequence, making sure that the parentheses are closed properly. Input: < [ < [ { < [ ] < { } > > } ] > { { ( ) } { < [ < > ] > }\nA: Let's think step by step.\nWe should process each input one by one and keep track of the stack configuration.\n0: empty stack\n1: < ; stack: <\n2: [ ; stack: < [\n3: < ; stack: < [ <\n4: [ ; stack: < [ < [\n5: { ; stack: < [ < [ {\n6: < ; stack: < [ < [ { <\n7: [ ; stack: < [ < [ { < [\n8: ] ; stack: < [ < [ { <\n9: < ; stack: < [ < [ { < <\n10: { ; stack: < [ < [ { < < {\n11: } ; stack: < [ < [ { < <\n12: > ; stack: < [ < [ { <\n13: > ; stack: < [ < [ {\n14: } ; stack: < [ < [\n15: ] ; stack: < [ <\n16: > ; stack: < [\n17: { ; stack: < [ {\n18: { ; stack: < [ { {\n19: ( ; stack: < [ { { (\n20: ) ; stack: < [ { {\n21: } ; stack: < [ {\n22: { ; stack: < [ { {\n23: < ; stack: < [ { { <\n24: [ ; stack: < [ { { < [\n25: < ; stack: < [ { { < [ <\n26: > ; stack: < [ { { < [\n27: ] ; stack: < [ { { <\n28: > ; stack: < [ { {\n29: } ; stack: < [ {\nNow, we have reached the end. The final stack is \"< [ {\".\nWe will need to pop out \"{\", \"[\", \"<\" one by one in that order.\nSo, we need \"}\", \"]\", \">\". So the answer is } ] >.Q: {{input}}\nA: Let's think step by step.\n" "doc_to_text": " Complete the rest of the sequence, making sure that the parentheses are closed properly. Input: [ { [\nA: Let's think step by step.\nWe should process each input one by one and keep track of the stack configuration.\n0: empty stack\n1: [ ; stack: [\n2: { ; stack: [ {\n3: [ ; stack: [ { [\nNow, we have reached the end. The final stack is \"[ { [\".\nWe will need to pop out \"[\", \"{\", \"[\" one by one in that order.\nSo, we need \"]\", \"}\", \"]\". So the answer is ] } ]. Complete the rest of the sequence, making sure that the parentheses are closed properly. Input: < > ( ( [ [ ( { } ) [ < > ] ]\nA: Let's think step by step.\nWe should process each input one by one and keep track of the stack configuration.\n0: empty stack\n1: < ; stack: <\n2: > ; stack: empty\n3: ( ; stack: (\n4: ( ; stack: ( (\n5: [ ; stack: ( ( [\n6: [ ; stack: ( ( [ [\n7: ( ; stack: ( ( [ [ (\n8: { ; stack: ( ( [ [ ( {\n9: } ; stack: ( ( [ [ (\n10: ) ; stack: ( ( [ [\n11: [ ; stack: ( ( [ [ [\n12: < ; stack: ( ( [ [ [ <\n13: > ; stack: ( ( [ [ [\n14: ] ; stack: ( ( [ [\n15: ] ; stack: ( ( [\nNow, we have reached the end. The final stack is \"( ( [\".\nWe will need to pop out \"[\", \"(\", \"(\" one by one in that order.\nSo, we need \"]\", \")\", \")\". So the answer is ] ) ). Complete the rest of the sequence, making sure that the parentheses are closed properly. Input: < [ < [ { < [ ] < { } > > } ] > { { ( ) } { < [ < > ] > }\nA: Let's think step by step.\nWe should process each input one by one and keep track of the stack configuration.\n0: empty stack\n1: < ; stack: <\n2: [ ; stack: < [\n3: < ; stack: < [ <\n4: [ ; stack: < [ < [\n5: { ; stack: < [ < [ {\n6: < ; stack: < [ < [ { <\n7: [ ; stack: < [ < [ { < [\n8: ] ; stack: < [ < [ { <\n9: < ; stack: < [ < [ { < <\n10: { ; stack: < [ < [ { < < {\n11: } ; stack: < [ < [ { < <\n12: > ; stack: < [ < [ { <\n13: > ; stack: < [ < [ {\n14: } ; stack: < [ < [\n15: ] ; stack: < [ <\n16: > ; stack: < [\n17: { ; stack: < [ {\n18: { ; stack: < [ { {\n19: ( ; stack: < [ { { (\n20: ) ; stack: < [ { {\n21: } ; stack: < [ {\n22: { ; stack: < [ { {\n23: < ; stack: < [ { { <\n24: [ ; stack: < [ { { < [\n25: < ; stack: < [ { { < [ <\n26: > ; stack: < [ { { < [\n27: ] ; stack: < [ { { <\n28: > ; stack: < [ { {\n29: } ; stack: < [ {\nNow, we have reached the end. The final stack is \"< [ {\".\nWe will need to pop out \"{\", \"[\", \"<\" one by one in that order.\nSo, we need \"}\", \"]\", \">\". So the answer is } ] >.Q: {{input}}\nA: Let's think step by step.\n"
"include": "_flan_cot_fewshot_template_yaml" "include": "_cot_fewshot_template_yaml"
"task": "bbh_flan_cot_fewshot_dyck_languages" "task": "bbh_cot_fewshot_dyck_languages"
lm_eval/tasks/bbh/flan_cot_fewshot/formal_fallacies.yaml → lm_eval/tasks/bbh/cot_fewshot/formal_fallacies.yaml
View file @ 5d3bf2e7
"dataset_name": "formal_fallacies" "dataset_name": "formal_fallacies"
"description": "Distinguish deductively valid arguments from formal fallacies.\n\n" "description": "Distinguish deductively valid arguments from formal fallacies.\n\n"
"doc_to_text": " \"It is not always easy to see who is related to whom -- and in which ways. The following argument pertains to this question: To begin with, Lesley is a close friend of Fernando. Moreover, being a close friend of Fernando or a schoolmate of Lowell is sufficient for being a great-grandfather of Leroy. It follows that Lesley is a great-grandfather of Leroy.\"\nIs the argument, given the explicitly stated premises, deductively valid or invalid?\nOptions:\n- valid\n- invalid\nA: Let's think step by step.\n(1) Lesley is a close friend of Fernando: Lesley = friend(Fernando).\n(2) Being a close friend of Fernando or a schoolmate of Lowell is sufficient for being a great-grandfather of Leroy: If X = friend(Fernando) OR SCHOOLMATE(Lowell), then X = great-grandfather(Leroy).\nHypothesis: Does it follow that Lesley is a great-grandfather of Leroy: Lesley = great-grandfather(Leroy)?\nLet’s see whether the Hypothesis can be deduced from the arguments (1) and (2) by logical reasoning?\nBy (1), we have Lesley = friend(Fernando). By (2), we have if Lesley = friend(Fernando), then Lesley = great-grandfather(Leroy).\nSo, it is true that Lesley is a great-grandfather of Leroy. So the answer is valid. \"It is not always easy to see who is related to whom -- and in which ways. The following argument pertains to this question: Whoever is not a great-grandfather of Clyde is a stepbrother of Brian. Being an ancestor of Dana is sufficient for not being a great-grandfather of Clyde. We may conclude: Everyone who is an ancestor of Dana is a stepbrother of Brian, too.\"\nIs the argument, given the explicitly stated premises, deductively valid or invalid?\nOptions:\n- valid\n- invalid\nA: Let's think step by step.\n(1) Whoever is not a great-grandfather of Clyde is a stepbrother of Brian: If X = NOT (great-grandfather(Clyde)), then X = stepbrother(Brian).\n(2): Being an ancestor of Dana is sufficient for not being a great-grandfather of Clyde: If X = ancestor(Dana), X = NOT (great-grandfather(Clyde)).\nHypothesis: Does it follow that everyone who is an ancestor of Dana is a stepbrother of Brian, too: If X = ancestor(Dana), then X = stepbrother(Brian)?\nLet’s see whether the Hypothesis can be deduced from the arguments (1) and (2) by logical reasoning?\nBy (2), we have if X = ancestor(Dana), X = NOT (great-grandfather(Clyde)).\nFurthermore, by (1), we have if X = NOT (great-grandfather(Clyde)), then X = stepbrother(Brian).\nBy the transitive relation rule in first-order logic, we then have: if X = ancestor(Dana), then X = stepbrother(Brian).\nSo, it is true that everyone who is an ancestor of Dana is a stepbrother of Brian. So the answer is valid. \"It is not always easy to grasp who is consuming which products. The following argument pertains to this question: Every infrequent user of Paul Mitchell shampoo is either a rare consumer of Nioxin shampoo or a loyal buyer of Caress soap, or both. No regular consumer of Lush soap is a rare consumer of Nioxin shampoo and, in the same time, a loyal buyer of Caress soap. It follows that whoever is an infrequent user of Paul Mitchell shampoo is not a regular consumer of Lush soap.\"\nIs the argument, given the explicitly stated premises, deductively valid or invalid?\nOptions:\n- valid\n- invalid\nA: Let's think step by step.\n(1) Every infrequent user of Paul Mitchell shampoo is either a rare consumer of Nioxin shampoo or a loyal buyer of Caress soap, or both: If X = infrequent-user(Paul Mitchell), then X = rare-consumer(Nioxin) OR X = loyal-buyer(Caress).\n(2): No regular consumer of Lush soap is a rare consumer of Nioxin shampoo and a loyal buyer of Caress soap at the same time. If X = regular-consumer(Lush), then X = NOT (rare-consumer(Nioxin) AND loyal-buyer(Caress)).\nHypothesis: Does it follow that whoever is an infrequent user of Paul Mitchell shampoo is not a regular consumer of Lush soap: If X = infrequent-user(Paul Mitchell), then X = NOT (regular-consumer(Lush))?\nLet’s see whether the Hypothesis can be deduced from the arguments (1) and (2) by logical reasoning?\nBy (1), we have if X = infrequent-user(Paul Mitchell), then X = rare-consumer(Nioxin) OR X = loyal-buyer(Caress). We need to consider both cases separately:\nThe case X = rare-consumer(Nioxin) does not appear in (2).\nThe case X = loyal-buyer(Caress) does not appear in (2), either.\nSo, from (1) and (2), we cannot necessarily deduce the Hypothesis. So the answer is invalid.Q: {{input}}\nA: Let's think step by step.\n" "doc_to_text": " \"It is not always easy to see who is related to whom -- and in which ways. The following argument pertains to this question: To begin with, Lesley is a close friend of Fernando. Moreover, being a close friend of Fernando or a schoolmate of Lowell is sufficient for being a great-grandfather of Leroy. It follows that Lesley is a great-grandfather of Leroy.\"\nIs the argument, given the explicitly stated premises, deductively valid or invalid?\nOptions:\n- valid\n- invalid\nA: Let's think step by step.\n(1) Lesley is a close friend of Fernando: Lesley = friend(Fernando).\n(2) Being a close friend of Fernando or a schoolmate of Lowell is sufficient for being a great-grandfather of Leroy: If X = friend(Fernando) OR SCHOOLMATE(Lowell), then X = great-grandfather(Leroy).\nHypothesis: Does it follow that Lesley is a great-grandfather of Leroy: Lesley = great-grandfather(Leroy)?\nLet’s see whether the Hypothesis can be deduced from the arguments (1) and (2) by logical reasoning?\nBy (1), we have Lesley = friend(Fernando). By (2), we have if Lesley = friend(Fernando), then Lesley = great-grandfather(Leroy).\nSo, it is true that Lesley is a great-grandfather of Leroy. So the answer is valid. \"It is not always easy to see who is related to whom -- and in which ways. The following argument pertains to this question: Whoever is not a great-grandfather of Clyde is a stepbrother of Brian. Being an ancestor of Dana is sufficient for not being a great-grandfather of Clyde. We may conclude: Everyone who is an ancestor of Dana is a stepbrother of Brian, too.\"\nIs the argument, given the explicitly stated premises, deductively valid or invalid?\nOptions:\n- valid\n- invalid\nA: Let's think step by step.\n(1) Whoever is not a great-grandfather of Clyde is a stepbrother of Brian: If X = NOT (great-grandfather(Clyde)), then X = stepbrother(Brian).\n(2): Being an ancestor of Dana is sufficient for not being a great-grandfather of Clyde: If X = ancestor(Dana), X = NOT (great-grandfather(Clyde)).\nHypothesis: Does it follow that everyone who is an ancestor of Dana is a stepbrother of Brian, too: If X = ancestor(Dana), then X = stepbrother(Brian)?\nLet’s see whether the Hypothesis can be deduced from the arguments (1) and (2) by logical reasoning?\nBy (2), we have if X = ancestor(Dana), X = NOT (great-grandfather(Clyde)).\nFurthermore, by (1), we have if X = NOT (great-grandfather(Clyde)), then X = stepbrother(Brian).\nBy the transitive relation rule in first-order logic, we then have: if X = ancestor(Dana), then X = stepbrother(Brian).\nSo, it is true that everyone who is an ancestor of Dana is a stepbrother of Brian. So the answer is valid. \"It is not always easy to grasp who is consuming which products. The following argument pertains to this question: Every infrequent user of Paul Mitchell shampoo is either a rare consumer of Nioxin shampoo or a loyal buyer of Caress soap, or both. No regular consumer of Lush soap is a rare consumer of Nioxin shampoo and, in the same time, a loyal buyer of Caress soap. It follows that whoever is an infrequent user of Paul Mitchell shampoo is not a regular consumer of Lush soap.\"\nIs the argument, given the explicitly stated premises, deductively valid or invalid?\nOptions:\n- valid\n- invalid\nA: Let's think step by step.\n(1) Every infrequent user of Paul Mitchell shampoo is either a rare consumer of Nioxin shampoo or a loyal buyer of Caress soap, or both: If X = infrequent-user(Paul Mitchell), then X = rare-consumer(Nioxin) OR X = loyal-buyer(Caress).\n(2): No regular consumer of Lush soap is a rare consumer of Nioxin shampoo and a loyal buyer of Caress soap at the same time. If X = regular-consumer(Lush), then X = NOT (rare-consumer(Nioxin) AND loyal-buyer(Caress)).\nHypothesis: Does it follow that whoever is an infrequent user of Paul Mitchell shampoo is not a regular consumer of Lush soap: If X = infrequent-user(Paul Mitchell), then X = NOT (regular-consumer(Lush))?\nLet’s see whether the Hypothesis can be deduced from the arguments (1) and (2) by logical reasoning?\nBy (1), we have if X = infrequent-user(Paul Mitchell), then X = rare-consumer(Nioxin) OR X = loyal-buyer(Caress). We need to consider both cases separately:\nThe case X = rare-consumer(Nioxin) does not appear in (2).\nThe case X = loyal-buyer(Caress) does not appear in (2), either.\nSo, from (1) and (2), we cannot necessarily deduce the Hypothesis. So the answer is invalid.Q: {{input}}\nA: Let's think step by step.\n"
"include": "_flan_cot_fewshot_template_yaml" "include": "_cot_fewshot_template_yaml"
"task": "bbh_flan_cot_fewshot_formal_fallacies" "task": "bbh_cot_fewshot_formal_fallacies"
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment