菜单
Structured prompt template for recovering the instruction that transformed one text into another, with labeled original/rewritten fields and explicit task framing
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Memory-efficient Keras generator that streams sharded CSV files in chunks, renders strokes to images on-the-fly, and yields batches for training on datasets too large for memory
Stack 1D convolutions for local feature extraction before bidirectional LSTMs to classify variable-length stroke sequences into hundreds of doodle categories
Partition a large dataset into N balanced shards using integer key modulo arithmetic for reproducible, class-interleaved splits across CSV files
Normalize raw stroke coordinates to 0-255 range, resample at uniform arc-length spacing, then apply Ramer-Douglas-Peucker simplification
Render stroke sequences to grayscale images with temporal intensity encoding where earlier strokes are brighter and later strokes fade to encode drawing order
Score each face in a video as anomalous by computing L2 distance from the embedding centroid of all faces, then convert to probability via logistic function
| name | llm-inverse-task-prompt-template |
| description | Structured prompt template for recovering the instruction that transformed one text into another, with labeled original/rewritten fields and explicit task framing |
To recover an unknown prompt from an (original, rewritten) text pair, frame the task explicitly: show the LLM both texts with clear labels, state that one was derived from the other via a prompt, and ask it to infer that prompt. The template structure — labeled fields, task description, output format constraint — is critical for consistent results across different LLMs.
TEMPLATE = """Below, the `Original Text` has been rewritten into `Rewritten Text`
by an LLM with a certain prompt/instruction. Analyze the differences
and infer the specific prompt that was used.
Original Text:
{original_text}
Rewritten Text:
{rewritten_text}
The prompt was:"""
def build_recovery_prompt(original, rewritten):
return TEMPLATE.format(
original_text=original[:1000],
rewritten_text=rewritten[:1000])
prompt = build_recovery_prompt(row['original_text'], row['rewritten_text'])
output = model.generate(tokenizer(prompt, return_tensors='pt').to(device),
max_new_tokens=50, do_sample=False)
recovered = tokenizer.decode(output[0], skip_special_tokens=True)
recovered = recovered.split("The prompt was:")[-1].strip()
Original Text:, Rewritten Text:)The prompt was:) to constrain output formatThe prompt was: forces concise output without preamble