Dataset Viewer
Auto-converted to Parquet Duplicate
name
stringclasses
8 values
instruction
stringclasses
1 value
description
stringclasses
1 value
keywords
listlengths
4
4
schema_version
stringclasses
1 value
config
unknown
metadata
unknown
files
listlengths
8
8
location
stringclasses
8 values
raman-lab/graphene-s01
You are given a Raman spectrum of a graphene sample in `/app/spectrum.csv`. It has a header row and two columns: `raman_shift_cm-1` (Raman shift in cm⁻¹) and `intensity` (counts). Fit a single Lorentzian plus a constant offset to each of the D (~1350 cm⁻¹), G (~1580 cm⁻¹) and 2D (~2690 cm⁻¹) peaks, using the model ...
Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.
[ "fitting", "physics", "scientific-computing", "raman" ]
1.1
{ "schema_version": "1.1", "artifacts": [], "task": { "name": "raman-lab/graphene-s01", "description": "Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.", "keywords": [ "fitting", "physics", "scientific-computing"...
{ "adapted_from": "terminal-bench/raman-fitting@2fd12b8", "split": "dev", "difficulty": "medium", "category": "scientific-computing" }
[ "environment/spectrum.csv", "instruction.md", "solution/solve.sh", "task.toml", "tests/data/reference.json", "tests/data/spectrum.csv", "tests/score.py", "tests/test.sh" ]
graphene-s01
raman-lab/graphene-s02
You are given a Raman spectrum of a graphene sample in `/app/spectrum.csv`. It has a header row and two columns: `raman_shift_cm-1` (Raman shift in cm⁻¹) and `intensity` (counts). Fit a single Lorentzian plus a constant offset to each of the D (~1350 cm⁻¹), G (~1580 cm⁻¹) and 2D (~2690 cm⁻¹) peaks, using the model ...
Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.
[ "fitting", "physics", "scientific-computing", "raman" ]
1.1
{ "schema_version": "1.1", "artifacts": [], "task": { "name": "raman-lab/graphene-s02", "description": "Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.", "keywords": [ "fitting", "physics", "scientific-computing"...
{ "adapted_from": "terminal-bench/raman-fitting@2fd12b8", "split": "dev", "difficulty": "medium", "category": "scientific-computing" }
[ "environment/spectrum.csv", "instruction.md", "solution/solve.sh", "task.toml", "tests/data/reference.json", "tests/data/spectrum.csv", "tests/score.py", "tests/test.sh" ]
graphene-s02
raman-lab/graphene-s03
You are given a Raman spectrum of a graphene sample in `/app/spectrum.csv`. It has a header row and two columns: `raman_shift_cm-1` (Raman shift in cm⁻¹) and `intensity` (counts). Fit a single Lorentzian plus a constant offset to each of the D (~1350 cm⁻¹), G (~1580 cm⁻¹) and 2D (~2690 cm⁻¹) peaks, using the model ...
Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.
[ "fitting", "physics", "scientific-computing", "raman" ]
1.1
{ "schema_version": "1.1", "artifacts": [], "task": { "name": "raman-lab/graphene-s03", "description": "Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.", "keywords": [ "fitting", "physics", "scientific-computing"...
{ "adapted_from": "terminal-bench/raman-fitting@2fd12b8", "split": "dev", "difficulty": "medium", "category": "scientific-computing" }
[ "environment/spectrum.csv", "instruction.md", "solution/solve.sh", "task.toml", "tests/data/reference.json", "tests/data/spectrum.csv", "tests/score.py", "tests/test.sh" ]
graphene-s03
raman-lab/graphene-s04
You are given a Raman spectrum of a graphene sample in `/app/spectrum.csv`. It has a header row and two columns: `raman_shift_cm-1` (Raman shift in cm⁻¹) and `intensity` (counts). Fit a single Lorentzian plus a constant offset to each of the D (~1350 cm⁻¹), G (~1580 cm⁻¹) and 2D (~2690 cm⁻¹) peaks, using the model ...
Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.
[ "fitting", "physics", "scientific-computing", "raman" ]
1.1
{ "schema_version": "1.1", "artifacts": [], "task": { "name": "raman-lab/graphene-s04", "description": "Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.", "keywords": [ "fitting", "physics", "scientific-computing"...
{ "adapted_from": "terminal-bench/raman-fitting@2fd12b8", "split": "dev", "difficulty": "medium", "category": "scientific-computing" }
[ "environment/spectrum.csv", "instruction.md", "solution/solve.sh", "task.toml", "tests/data/reference.json", "tests/data/spectrum.csv", "tests/score.py", "tests/test.sh" ]
graphene-s04
raman-lab/graphene-s05
You are given a Raman spectrum of a graphene sample in `/app/spectrum.csv`. It has a header row and two columns: `raman_shift_cm-1` (Raman shift in cm⁻¹) and `intensity` (counts). Fit a single Lorentzian plus a constant offset to each of the D (~1350 cm⁻¹), G (~1580 cm⁻¹) and 2D (~2690 cm⁻¹) peaks, using the model ...
Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.
[ "fitting", "physics", "scientific-computing", "raman" ]
1.1
{ "schema_version": "1.1", "artifacts": [], "task": { "name": "raman-lab/graphene-s05", "description": "Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.", "keywords": [ "fitting", "physics", "scientific-computing"...
{ "adapted_from": "terminal-bench/raman-fitting@2fd12b8", "split": "test", "difficulty": "medium", "category": "scientific-computing" }
[ "environment/spectrum.csv", "instruction.md", "solution/solve.sh", "task.toml", "tests/data/reference.json", "tests/data/spectrum.csv", "tests/score.py", "tests/test.sh" ]
graphene-s05
raman-lab/graphene-s06
You are given a Raman spectrum of a graphene sample in `/app/spectrum.csv`. It has a header row and two columns: `raman_shift_cm-1` (Raman shift in cm⁻¹) and `intensity` (counts). Fit a single Lorentzian plus a constant offset to each of the D (~1350 cm⁻¹), G (~1580 cm⁻¹) and 2D (~2690 cm⁻¹) peaks, using the model ...
Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.
[ "fitting", "physics", "scientific-computing", "raman" ]
1.1
{ "schema_version": "1.1", "artifacts": [], "task": { "name": "raman-lab/graphene-s06", "description": "Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.", "keywords": [ "fitting", "physics", "scientific-computing"...
{ "adapted_from": "terminal-bench/raman-fitting@2fd12b8", "split": "test", "difficulty": "medium", "category": "scientific-computing" }
[ "environment/spectrum.csv", "instruction.md", "solution/solve.sh", "task.toml", "tests/data/reference.json", "tests/data/spectrum.csv", "tests/score.py", "tests/test.sh" ]
graphene-s06
raman-lab/graphene-s07
You are given a Raman spectrum of a graphene sample in `/app/spectrum.csv`. It has a header row and two columns: `raman_shift_cm-1` (Raman shift in cm⁻¹) and `intensity` (counts). Fit a single Lorentzian plus a constant offset to each of the D (~1350 cm⁻¹), G (~1580 cm⁻¹) and 2D (~2690 cm⁻¹) peaks, using the model ...
Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.
[ "fitting", "physics", "scientific-computing", "raman" ]
1.1
{ "schema_version": "1.1", "artifacts": [], "task": { "name": "raman-lab/graphene-s07", "description": "Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.", "keywords": [ "fitting", "physics", "scientific-computing"...
{ "adapted_from": "terminal-bench/raman-fitting@2fd12b8", "split": "test", "difficulty": "medium", "category": "scientific-computing" }
[ "environment/spectrum.csv", "instruction.md", "solution/solve.sh", "task.toml", "tests/data/reference.json", "tests/data/spectrum.csv", "tests/score.py", "tests/test.sh" ]
graphene-s07
raman-lab/graphene-s08
You are given a Raman spectrum of a graphene sample in `/app/spectrum.csv`. It has a header row and two columns: `raman_shift_cm-1` (Raman shift in cm⁻¹) and `intensity` (counts). Fit a single Lorentzian plus a constant offset to each of the D (~1350 cm⁻¹), G (~1580 cm⁻¹) and 2D (~2690 cm⁻¹) peaks, using the model ...
Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.
[ "fitting", "physics", "scientific-computing", "raman" ]
1.1
{ "schema_version": "1.1", "artifacts": [], "task": { "name": "raman-lab/graphene-s08", "description": "Fit D, G and 2D Lorentzians to a graphene Raman spectrum and report I_D/I_G. Adapted from terminal-bench/raman-fitting.", "keywords": [ "fitting", "physics", "scientific-computing"...
{ "adapted_from": "terminal-bench/raman-fitting@2fd12b8", "split": "test", "difficulty": "medium", "category": "scientific-computing" }
[ "environment/spectrum.csv", "instruction.md", "solution/solve.sh", "task.toml", "tests/data/reference.json", "tests/data/spectrum.csv", "tests/score.py", "tests/test.sh" ]
graphene-s08

raman-lab

Eight Harbor tasks for fitting graphene Raman spectra, adapted from Terminal-Bench 2's raman-fitting. Each task gives the agent one synthetic spectrum with known parameters. The agent must fit Lorentzians to the D, G and 2D peaks and report the D-to-G intensity ratio. The spectra vary in noise and defect level.

This is a development set, used in Public Benchmarks won't save you! How to build better agentic evals. These are the exact task versions behind the 2026-11-18 run: their content digests match the lock.json files in burtenshaw/raman-lab-runs.

Each task folder holds:

  • instruction.md: the task given to the agent
  • environment/: the container definition and the spectrum
  • solution/: the reference fit
  • tests/: the verifier, with reference values and tolerances
  • task.toml: resources, timeouts and metadata

Run it with Harbor:

hf download burtenshaw/raman-lab --repo-type dataset --local-dir raman-lab
harbor run -p raman-lab -a oracle -e hf-sandbox

The generator, analysis code and results are in burtenshaw/raman-lab-evals.

Downloads last month
-