instance_id stringlengths 47 52 | resolved bool 2
classes | messages listlengths 14 184 |
|---|---|---|
coveragepy__94b231f7b6175a38e97ff2555a4553bfa0b58389 | false | [
{
"content": "You are OpenHands agent, a helpful AI assistant that can interact with a computer to solve tasks.\n\n<ROLE>\nYour primary role is to assist users by executing commands, modifying code, and solving technical problems effectively. You should be thorough, methodical, and prioritize quality over speed... |
aiohttp__1e7faf857c9c68400848ba43a894479681736b0c | false | [{"content":"You are OpenHands agent, a helpful AI assistant that can interact with a computer to so(...TRUNCATED) |
aiohttp__f4b5552c49a43359d3e2a14dc7d0d44262f9e41f | false | [{"content":"You are OpenHands agent, a helpful AI assistant that can interact with a computer to so(...TRUNCATED) |
aiohttp__be27f7926f0da9b4129d98cf628eea99dc9e3411 | false | [{"content":"You are OpenHands agent, a helpful AI assistant that can interact with a computer to so(...TRUNCATED) |
numpy__61d2a445881f80b52bc9facdbd4f58f6e74c637b | false | [{"content":"You are OpenHands agent, a helpful AI assistant that can interact with a computer to so(...TRUNCATED) |
numpy__0a00dc9d5a6c7ad87a8117bace5a6775b0d737df | true | [{"content":"You are OpenHands agent, a helpful AI assistant that can interact with a computer to so(...TRUNCATED) |
coveragepy__694435b844760ca0443cf87bf73a537f06c37d7d | false | [{"content":"You are OpenHands agent, a helpful AI assistant that can interact with a computer to so(...TRUNCATED) |
numpy__f7f87596aee84daba42d90736af2eeb740eb14f7 | false | [{"content":"You are OpenHands agent, a helpful AI assistant that can interact with a computer to so(...TRUNCATED) |
pandas__f82b1c6329a1c2de20e453e6bae5fbe0beb69360 | false | [{"content":"You are OpenHands agent, a helpful AI assistant that can interact with a computer to so(...TRUNCATED) |
pandas__39b7a6d5614dc2a3e0ce9d799fa1dc98c1e3cfd5 | false | [{"content":"You are OpenHands agent, a helpful AI assistant that can interact with a computer to so(...TRUNCATED) |
r2egym_qwen3next80b_1500i_text2x
Verbosity-ablation variant of synthetic-code-training/r2egym_qwen3next80b_1500i: the prose before every tool call is rewritten by nvidia/deepseek-ai/deepseek-v4-flash to 2 times its own length in Qwen3 tokens (accepted band ±25 %, up to 3 rounds). Of 30493 turns, 182 were not rephrased (empty prose, or a target outside 4-2000 tokens) and 760 missed the band; both keep their original prose.
Construction (shared by the _text0x/0.5x/2x/4x/8x siblings): think and task_tracker turns are kept
verbatim, as are the system prompt, task, tool calls and tool results. For the rephrased siblings the
rephraser saw only the current turn (its prose + its tool call) and wrote the four versions in one response
as a ladder (0.5x shortens the original, 2x elaborates it, 4x elaborates the 2x, 8x elaborates the 4x), so
the four lengths share one meaning. A turn's targets are multiples of its own prose length, so empty turns
stay empty.
Malformed tool calls (garbled <tool_call> JSON / <invoke>) are kept verbatim as the call part.
| value | |
|---|---|
| rows | 1500 |
| assistant turns (excluding think/task_tracker) | 30493 |
| think/task_tracker turns kept verbatim | 8 |
| prose tokens per turn: mean / median | 134.5 / 41 |
| prose tokens / base prose tokens, rephrased turns: mean / median | 1.91 / 1.88 |
| turns not rephrased (empty prose or target outside 4-2000 tokens) | 182 |
| turns with empty prose | 85 |
| turns kept original prose (band missed) | 760 |
Built with tools/verbosity_rephrase in the benchmarks repo.
- Downloads last month
- 39