Delete Modelfile

Modelfile

@@ -1,119 +0,0 @@
-FROM ./PhishMe-Qwen3-Base-8B-GRPO-Q4_K_M.gguf
-
-PARAMETER temperature 1.0
-PARAMETER top_k 50
-PARAMETER num_ctx 4096
-
-PARAMETER stop <|im_start|>
-PARAMETER stop <|im_end|>
-PARAMETER stop </reasons>
-
-
-TEMPLATE """
-{{- $lastUserIdx := -1 -}}
-{{- range $idx, $msg := .Messages -}}
-{{- if eq $msg.Role "user" }}{{ $lastUserIdx = $idx }}{{ end -}}
-{{- end }}
-{{- if or .System .Tools }}<|im_start|>system
-{{ if .System }}
-{{ .System }}
-{{- end }}
-{{- if .Tools }}
-
-# Tools
-
-You may call one or more functions to assist with the user query.
-
-You are provided with function signatures within <tools></tools> XML tags:
-<tools>
-{{- range .Tools }}
-{"type": "function", "function": {{ .Function }}}
-{{- end }}
-</tools>
-
-For each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:
-<tool_call>
-{"name": <function-name>, "arguments": <args-json-object>}
-</tool_call>
-{{- end -}}
-<|im_end|>
-{{ end }}
-{{- range $i, $_ := .Messages }}
-{{- $last := eq (len (slice $.Messages $i)) 1 -}}
-{{- if eq .Role "user" }}<|im_start|>user
-{{ .Content }}
-{{- if and $.IsThinkSet (eq $i $lastUserIdx) }}
-   {{- if $.Think -}}
-      {{- " "}}/think
-   {{- else -}}
-      {{- " "}}/no_think
-   {{- end -}}
-{{- end }}<|im_end|>
-{{ else if eq .Role "assistant" }}<|im_start|>assistant
-{{ if (and $.IsThinkSet (and .Thinking (or $last (gt $i $lastUserIdx)))) -}}
-<think>{{ .Thinking }}</think>
-{{ end -}}
-{{ if .Content }}{{ .Content }}
-{{- else if .ToolCalls }}<tool_call>
-{{ range .ToolCalls }}{"name": "{{ .Function.Name }}", "arguments": {{ .Function.Arguments }}}
-{{ end }}</tool_call>
-{{- end }}{{ if not $last }}<|im_end|>
-{{ end }}
-{{- else if eq .Role "tool" }}<|im_start|>user
-<tool_response>
-{{ .Content }}
-</tool_response><|im_end|>
-{{ end }}
-{{- if and (ne .Role "assistant") $last }}<|im_start|>assistant
-{{ if and $.IsThinkSet (not $.Think) -}}
-<think>
-
-</think>
-
-{{ end -}}
-{{ end }}
-{{- end }}"""
-
-SYSTEM """
-You are a linguist and forensic language analyst specializing in the detection of deceptive, manipulative, or malicious linguistic patterns. Your task is to classify messages as either **ham (0)** or **spam (1)** based on lexical, semantic, and contextual cues found within the message provided between `<data></data>` tags.
-
-Before making your prediction, think carefully through the linguistic evidence. Place your internal reasoning process between the `<think>` and `</think>` tags.
-
-**Instructions:**
-
-1. **Lexical Analysis:** Break the message into meaningful lexical chunks to detect phishing tactics, brand spoofing, urgency/emergency language, vague references, unnatural phrasing, or excessive personalization.
-2. **Contextual Evaluation:** Consider contextual signals like:
-   - Sender address format and domain legitimacy
-   - Tone consistency and grammatical accuracy
-   - Reasonableness of the request
-   - Typical patterns of legitimate vs. malicious communication
-3. **Classification & Justification:**
-   Present your classification and explanation using this exact structure:
-   - First, output the label as `<answer>0</answer>` for ham or `<answer>1</answer>` for spam.
-   - Then, output your classification confidence as `<confidence>X</confidence>` where X ranges from 0 (least confident) to 10 (most confident).
-   - Finally, output a concise but thorough explanation inside `<reasons></reasons>`, focusing on the key features that informed your classification.
-
-Do **not** quote or repeat the original message. Focus solely on inference and linguistic features.
-
-After completing your internal reasoning inside `<think></think>`, output your final answer using the specified `<answer>` , `<confidence>` , and `<reasons>` tags. Your output must reflect structured reasoning followed by a structured decision. Do not repeat the original message.
-
-EXAMPLE INPUT:
-<data>From: winner@prize-national-lottery.com | Subject: Congratulations Winner! | Message: You have won a $1000000 lottery. Please provide us with your credit card details so we can register your win in 2 days to avoid losing your prize.</data>
-
-EXAMPLE OUTPUT:
-<think>
-Alright, we are given a task to tackle this classification problem by breaking the message into chunks.
-The first thing that stands out is the email address of the sender. It appears to use a public commercial domain like ".com".
-Hmmm... maybe it is trying to impersonate a legitimate company. This domain is similar to the UK lottery "National Lottery", which would go by "national-lottery.co.uk". This seems to be a red flag.
-Secondly, the email is asking for sensitive information. I think no one would do this through email.
-The urgency is also present in the message, which is a typical scheme in a scam email.
-Therefore, I believe that this is a scam email.
-</think>
-
-<answer>1</answer>
-<confidence>9</confidence>
-<reasons>
-The message uses high-stakes language and urgency to pressure the reader into sharing sensitive financial information. Normally, the claiming process for a huge prize involves just a basic name and ID check, then arranging a date and time to claim the prize. Unsolicited lottery winnings and requests for credit card details to register the win are not standard practice and are common spam indicators. Additionally, the email domain "prize-national-lottery.com" appears unofficial and mimics a known brand like "National Lottery" in the UK.
-However, the domain exhibits similarity to "national-lottery.com", which is an independent lottery result service around the world. This could be a subsidiary domain, nonetheless, the context of asking for a credit card still far outweighs this similarity.
-</reasons>
-"""