Dataset Viewer
Auto-converted to Parquet Duplicate
source
stringclasses
2 values
prompt
stringlengths
1
8.93k
system
stringclasses
0 values
messages
listlengths
2
2
tools
stringclasses
63 values
deepseek-v4-distill-8000x
How does the inclusion of higher-order curvature terms in the Regge action affect the scaling behavior of the spectral dimension in causal dynamical triangulations, and what implications does this have for the emergence of semiclassical spacetime geometry?
null
[ { "role": "user", "content": "How does the inclusion of higher-order curvature terms in the Regge action affect the scaling behavior of the spectral dimension in causal dynamical triangulations, and what implications does this have for the emergence of semiclassical spacetime geometry?", "reasoning_cont...
[]
deepseek-v4-distill-8000x
Write a python 2 script that monitors RabbitMQ connections and channels to identify congested connections with high numbers of unacknowledged messages. The script should include functions to execute shell commands, parse channel and connection data from rabbitmqctl output using regular expressions, and identify connect...
null
[ { "role": "user", "content": "Write a python 2 script that monitors RabbitMQ connections and channels to identify congested connections with high numbers of unacknowledged messages. The script should include functions to execute shell commands, parse channel and connection data from rabbitmqctl output using...
[]
deepseek-v4-distill-8000x
Determine whether 467358 is divisible by 11. Explain the divisibility rule you used.
null
[ { "role": "user", "content": "Determine whether 467358 is divisible by 11. Explain the divisibility rule you used.", "reasoning_content": null, "tool_calls": "[]" }, { "role": "assistant", "content": "<think>\nWe need to determine if 467358 is divisible by 11. The divisibility rule for 1...
[]
fable-5-traces
…[earlier truncated]… loads(fn[\"arguments\"])\n except (json.JSONDecodeError, TypeError):\n args = fn[\"arguments\"]\n call = json.dumps({\"name\": fn[\"name\"], \"arguments\": args}, ensure_ascii=False)", "new_string": " for tc in expert.get(\"tool_calls\", []):\n fn = tc[\"function...
null
[ { "role": "user", "content": "…[earlier truncated]…\nloads(fn[\\\"arguments\\\"])\\n except (json.JSONDecodeError, TypeError):\\n args = fn[\\\"arguments\\\"]\\n call = json.dumps({\\\"name\\\": fn[\\\"name\\\"], \\\"arguments\\\": args}, ensure_ascii=False)\", \"new_string\": \" ...
[{"type": "function", "function": {"name": "Edit", "parameters": {"type": "object", "properties": {"file_path": {"type": "string"}, "new_string": {"type": "string"}, "old_string": {"type": "string"}, "replace_all": {"type": "boolean"}}, "additionalProperties": true, "required": ["file_path", "new_string", "old_string",...
fable-5-traces
…[earlier truncated]… "hp": 100, "state": "patrol", "pos": [ 29.7, 0, -19.3 ] } } ASSISTANT (tool call) Edit input={"replace_all": false, "file_path": "/home/lane/AIArchives/archives/neonstrike/test/playtest.mjs", "old_string": "// teleport next to bot 0 with clear LOS, re-aim before ev...
null
[ { "role": "user", "content": "…[earlier truncated]…\n \"hp\": 100,\n \"state\": \"patrol\",\n \"pos\": [\n 29.7,\n 0,\n -19.3\n ]\n }\n}\nASSISTANT (tool call) Edit input={\"replace_all\": false, \"file_path\": \"/home/lane/AIArchives/archives/neonstrike/test/playtest.mjs\", \"old_s...
[{"type": "function", "function": {"name": "Write", "parameters": {"type": "object", "properties": {"content": {"type": "string"}, "file_path": {"type": "string"}}, "additionalProperties": true, "required": ["content", "file_path"]}}}, {"type": "function", "function": {"name": "bash", "description": "Run shell commands...
deepseek-v4-distill-8000x
When there is a collision of 2 disc shaped galaxies, there is a tail formation created from both the galaxies. I read here that this was due to tidal forces, but I couldn't figure out how this happens. Is there a simple explanation as to why and how tidal tails are formed? \nAlso, are tidal tails in the same plane as t...
null
[ { "role": "user", "content": "When there is a collision of 2 disc shaped galaxies, there is a tail formation created from both the galaxies. I read here that this was due to tidal forces, but I couldn't figure out how this happens. Is there a simple explanation as to why and how tidal tails are formed? \\nA...
[]
deepseek-v4-distill-8000x
Below you are given code written in python that contains the function 'stuff', which manipulates a string, as well as an input to the function. Your job is to predict the output of 'stuff' given the provided input.\n\nHere is the code:\n```python\ndef stuff(a: str) -> str:\n from functools import lru_cache\n\n cl...
null
[ { "role": "user", "content": "Below you are given code written in python that contains the function 'stuff', which manipulates a string, as well as an input to the function. Your job is to predict the output of 'stuff' given the provided input.\\n\\nHere is the code:\\n```python\\ndef stuff(a: str) -> str:\...
[]
deepseek-v4-distill-8000x
What is the relationship between legitimacy and the nature of causation?
null
[ { "role": "user", "content": "What is the relationship between legitimacy and the nature of causation?", "reasoning_content": null, "tool_calls": "[]" }, { "role": "assistant", "content": "<think>\nThis is a complex philosophical question that touches on political theory and metaphysics....
[]
fable-5-traces
…[earlier truncated]… 4ffb74fb5\nexport MSYS_NO_PATHCONV=1\ntimeout 25 docker exec -i -e TERM=xterm-256color --user codex -w /workspace $C script -qfec \"stty cols 120 rows 30 2>/dev/null; exec claude --model claude-sonnet-4-6 --permission-mode bypassPermissions\" /dev/null > /tmp/claude_out.bin 2>&1; echo \"exit=$?\";...
null
[ { "role": "user", "content": "…[earlier truncated]…\n4ffb74fb5\\nexport MSYS_NO_PATHCONV=1\\ntimeout 25 docker exec -i -e TERM=xterm-256color --user codex -w /workspace $C script -qfec \\\"stty cols 120 rows 30 2>/dev/null; exec claude --model claude-sonnet-4-6 --permission-mode bypassPermissions\\\" /dev/n...
[{"type": "function", "function": {"name": "Grep", "parameters": {"type": "object", "properties": {"-A": {"type": "integer"}, "-B": {"type": "integer"}, "output_mode": {"type": "string"}, "path": {"type": "string"}, "pattern": {"type": "string"}}, "additionalProperties": true, "required": ["-A", "-B", "output_mode", "p...
deepseek-v4-distill-8000x
Write a python function that loads and parses an FTS3 configuration file, handling special cases like missing headers, array values, and database connection strings, while also mapping configuration options to a dictionary and processing roles and permissions.
null
[ { "role": "user", "content": "Write a python function that loads and parses an FTS3 configuration file, handling special cases like missing headers, array values, and database connection strings, while also mapping configuration options to a dictionary and processing roles and permissions.", "reasoning_...
[]
deepseek-v4-distill-8000x
In a Dyna-style architecture, how can uncertainty propagation through a learned latent dynamics model be incorporated into planning to guarantee asymptotic optimality under stochastic transitions, and what theoretical bounds characterize the required model fidelity?
null
[ { "role": "user", "content": "In a Dyna-style architecture, how can uncertainty propagation through a learned latent dynamics model be incorporated into planning to guarantee asymptotic optimality under stochastic transitions, and what theoretical bounds characterize the required model fidelity?", "reas...
[]
fable-5-traces
…[earlier truncated]… one = None, 3822 ) -> Path: 3823 if session_id is None: 3824 session_id = str(uuid.uuid4()) 3825 workspace_root, workspace = self._prepare_workspace(session_id, prompt_input, self.container_kind) 3826 home_dir = Path(tempfile.mkdtemp(prefix=f"{self.container...
null
[ { "role": "user", "content": "…[earlier truncated]…\none = None,\n3822\t ) -> Path:\n3823\t if session_id is None:\n3824\t session_id = str(uuid.uuid4())\n3825\t workspace_root, workspace = self._prepare_workspace(session_id, prompt_input, self.container_kind)\n3826\t home...
[{"type": "function", "function": {"name": "bash", "description": "Run shell commands in the workspace.", "parameters": {"type": "object", "properties": {"command": {"type": "string"}, "cmd": {"type": "string"}, "cwd": {"type": "string"}, "description": {"type": "string"}, "timeout": {"type": "integer"}}, "anyOf": [{"r...
fable-5-traces
…[earlier truncated]… in your context — no need to Read it back) ASSISTANT (tool call) Edit input={"replace_all": false, "file_path": "D:\\project-5\\js\\overworld.js", "old_string": " // map connections at edges\n const conn = m.connections ?? {};", "new_string": " // map connections at edges (blocked until ...
null
[ { "role": "user", "content": "…[earlier truncated]…\n in your context — no need to Read it back)\nASSISTANT (tool call) Edit input={\"replace_all\": false, \"file_path\": \"D:\\\\project-5\\\\js\\\\overworld.js\", \"old_string\": \" // map connections at edges\\n const conn = m.connections ?? {};\", \...
[{"type": "function", "function": {"name": "Edit", "parameters": {"type": "object", "properties": {"file_path": {"type": "string"}, "new_string": {"type": "string"}, "old_string": {"type": "string"}, "replace_all": {"type": "boolean"}}, "additionalProperties": true, "required": ["file_path", "new_string", "old_string",...
deepseek-v4-distill-8000x
Article: In the US, African Americans, who include multiracial people, earn 75% of what white people earn. In Brazil, people of color earn less than 50% of what whites earn. Some have posited that the facts of lower socioeconomic status for people of color suggest that Brazil practices a kind of one-drop rule, or discr...
null
[ { "role": "user", "content": "Article: In the US, African Americans, who include multiracial people, earn 75% of what white people earn. In Brazil, people of color earn less than 50% of what whites earn. Some have posited that the facts of lower socioeconomic status for people of color suggest that Brazil p...
[]
fable-5-traces
USER: You are building the multiplayer layer for NEONSTRIKE, a browser FPS at /home/lane/AIArchives/archives/neonstrike/. Create EXACTLY TWO NEW FILES and touch nothing else: 1. /home/lane/AIArchives/archives/neonstrike/app/server/index.js (CommonJS, Node) 2. /home/lane/AIArchives/archives/neonstrike/app/public/js/ne...
null
[ { "role": "user", "content": "USER: You are building the multiplayer layer for NEONSTRIKE, a browser FPS at /home/lane/AIArchives/archives/neonstrike/. Create EXACTLY TWO NEW FILES and touch nothing else:\n\n1. /home/lane/AIArchives/archives/neonstrike/app/server/index.js (CommonJS, Node)\n2. /home/lane/AI...
[{"type": "function", "function": {"name": "Read", "parameters": {"type": "object", "properties": {"file_path": {"type": "string"}}, "additionalProperties": true, "required": ["file_path"]}}}, {"type": "function", "function": {"name": "bash", "description": "Run shell commands in the workspace.", "parameters": {"type":...
End of preview. Expand in Data Studio

This dataset merges two sources into a single SFT-ready set with train/eval splits:

Because it incorporates AGPL-3.0-licensed data, this combined dataset is distributed under AGPL-3.0.

Split

Both source datasets were loaded in full (no weighting/subsampling), normalized to a common messages (OpenAI-style, tool calls preserved) schema, concatenated, filtered to drop rows over ~8192 tokens (estimated at 4 chars/token, not a real tokenizer), shuffled together (seed 42), and split 80% train / 20% eval on the combined pool.

Schema

  • source: origin dataset name
  • prompt: original user prompt/input text
  • system: system prompt, if any
  • messages: JSON-encoded list of chat messages (role/content/tool_calls/reasoning_content)
  • tools: JSON-encoded list of tool schemas available in that trace (empty for DeepSeek rows)

Process

This dataset was merged, split, and published using Claude Code.

Downloads last month
60

Models trained or fine-tuned on heyimmitzu/merged_reasoning_agent_dataset